Thematic Network

Road Safety and Environmental Benefit-Cost and Cost-Effectiveness Analysis for Use in Decision-Making

- WP 4 – Testing the efficiency assessment tools on selected road safety measures

Public

May 2005

Funded by the European Commission

- WP 4 – Testing the efficiency assessment tools on selected road safety measures

Public

ROSEBUD Road Safety and Environmental Benefit-Cost and Cost-Effectiveness Analysis for Use in Decision-Making

Contract No: GTC2/2000/33020

Network co-ordinator: Federal Highway Research Institute - BASt,

WP 4 co-ordinator: Austrian Road Safety Board – KfV,

Editors: Martin Winkelbauer and Christian Stefan (KfV)

Partners in WP 4: Centre d’Etudes Techniques de l’Equipement du Sud Quest – CETE SO, Technion, Transportation Research Institute – TRI, Israel National Technical University of Athens – NTUA, Greece Transport Research Centre – CDV, Technical Research Centre of Finland – VTT, Finland Austrian Road Safety Board – KfV, Austria

Report No: D6

Date: May 2005

Thematic Network funded by the European Commission, Directorate General for Energy and Transport responding the Thematic programme “Competitive and Sustainable Growth” of the 5th framework programme

TABLE OF CONTENTS

INTRODUCTION ...... 7 CASE A: ANTI-LOCK BRAKING SYSTEMS FOR MOTORCYCLES ...... 12 by Martin Winkelbauer, ...... 12 Austrian Road Safety Board (KfV), Austria ...... 12 CASE B1: SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL (, A22 MOTORWAY)...... 24 by Christian Stefan ...... 24 Austian Road Safety Board (KfV), Austria ...... 24 CASE B2: AUTOMATIC SPEED ENFORCEMENT ON THE A13 MOTORWAY (NL) .....44 by Christian Stefan ...... 44 Austian Road Safety Board (KfV), Austria ...... 44 CASE C1: DAYTIME RUNNING LIGHTS IN THE CZECH REPUBLIC ...... 53 by Petr Pokorný ...... 53 Transport Research Centre, CDV, The Czech Republic...... 53 CASE C2: DAYTIME RUNNING LIGHTS IN AUSTRIA...... 63 by Petr Pokorný ...... 63 Transport Research Centre, CDV, The Czech Republic...... 63 CASE E1: FOUR-ARM ROUNDABOUTS IN URBAN AREAS IN THE CZECH REPUBLIC...... 72 by Petr Pokorný ...... 72 Transport Research Centre, CDV, The Czech Republic...... 72 CASE E2: SPEED HUMPS ON LOCAL STREETS ...... 82 by Victoria Gitelman and Shalom Hakkert, ...... 82 Transportation Research Institute, Technion, Israel ...... 82 CASE E3: TRAFFIC CALMING MEASURES ...... 96 by George Yannis and Petros Evgenikos ...... 96 NTUA / DTPE, Greece...... 96 CASE F1: GRADE-SEPARATION AT RAILROAD CROSSINGS...... 114 by Marko Nokkala, ...... 114 VTT Building and Transport, Finland ...... 114 CASE F2: GRADE-SEPARATION AT ROAD-RAIL CROSSINGS...... 128 by Victoria Gitelman and Shalom Hakkert, ...... 128 Transportation Research Institute, Technion, Israel ...... 128 CASE G: MEASURE AGAINST COLLISIONS WITH TREES ...... 141 by Philippe Lejeune,...... 141 CETE SO, France...... 141 CASE H: INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION...... 155 by Victoria Gitelman and Shalom Hakkert, ...... 155 Transportation Research Institute, Technion, Israel ...... 155 CASE I1: INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL) ...... 168 by George Yannis and Eleonora Papadimitriou ...... 168 NTUA / DTPE, Greece...... 168 CASE I2: CONCENTRATED GENERAL ENFORCEMENT ON INTERURBAN ROADS IN ISRAEL ...... 185 by Victoria Gitelman and Shalom Hakkert, ...... 185 Transportation Research Institute, Technion, Israel ...... 185 CASE J1: 2 + 1 ROADS IN FINLAND ...... 204 by Marko Nokkala, ...... 204 VTT Building and Transport, Finland ...... 204 CASE J2: 2 + 1 ROADS IN SWEDEN ...... 214 by Marko Nokkala, ...... 214 VTT Building and Transport, Finland ...... 214 CASE K: COMPULSORY BICYCLE HELMET WEARING...... 222 by Martin Winkelbauer, ...... 222 Austrian Road Safety Board, KfV, Austria...... 222 SHORT TRAINING COURSE ON EFFICIENCY ASSESSMENT...... 241 CONCLUSIONS ...... 249 ANNEXES ...... 261 INTRODUCTION

INTRODUCTION by Victoria Gitelman and Martin Winkelbauer

Every year, more than 1 million injury accidents (including 50,000 fatalities and 1.7 million people injured) occur on public roads throughout the European Union. Hence, improving road safety was given top priority in the European Union’s Transport Policy. To reach the overall objective of halving the number of fatalities by 2010, it is essential to know the reduction potentials of the wide variety of already-existing road safety measures. A prerequisite for this task is reliable knowledge about the effectiveness and efficiency of the road safety measures considered. Previous ROSEBUD work packages answered the question how efficiency assessment tools are currently used in different countries (WP1), what factors prevent the use of those tools (WP2) and what can be done to overcome existing barriers and shortcomings (WP3). The main task of work package 4 (WP4) is to test the developed efficiency assessment tools on selected road safety measures. The WP4 program was as follows: • to carry out a certain number of Efficiency Assessment Studies • to report experiences gained from those studies • and to evaluate, through these practical examples, the results of previous work packages (in treating barriers and the use of standardised procedures, respectively)

1.1 Selecting the cases for Efficiency Assessment

In accordance with the above program, eleven test cases were chosen, covering as many types of road safety measures as possible (see Table 1). The applicability of the developed analyses techniques of WP3 were tested in light of both the limitation of available data and restrictions of decision-making procedures in different countries. Table 1: Selected cases for evaluation in work package 4 Nr. Case Study Road Safety Approach Level Countries Responsibility A ABS motorcycle Vehicle National AT AT B Section control User + Enforcement Local AT, NL AT C Daytime running lights Vehicle + User National AT, CZ CZ D Speed cameras1 User + Enforcement Local FI, IL IL E Traffic calming (urban areas) Infrastructure Local CZ, GR, IL IL F Railroad crossings Infrastructure Local FI, IL FI Measures against collisions Local + G Infrastructure FR FR with trees (guardrails) National Road improvement mix (rural Local + H Infrastructure IL IL areas, national network) National Intensive police enforcement I User + Enforcement National GR, IL GR (speed and alcohol) J 2+1 roads Infrastructure Regional FI, SW FI Compulsory helmet regulation K User National AT, DE AT for cyclists

1 At the workshop in Bordeaux in December 2004, WP4 members decided to cancel the whole of case D due to missing data on the topic. Furthermore, speed enforcement is covered quite extensively by two other cases - case study B and I.

Page 7 INTRODUCTION

Selected cases were carried out by several working groups consisting of one to three WP4 members and a non-specified number of URG members. Various considerations were taken into account before a safety measure was considered a test case. They are: 1. Different categories of safety-related measures as defined by WP1, i.e. user-related measures, vehicle-related measures, infrastructure related measures, organisation and rescue services. The available experiences and data from different countries have been analysed with the purpose to cover as many safety-related categories as possible. 2. Safety measures can be attributed to different levels of implementation (national, regional and local) which influences the effect of the treatment on its environment. Local measures are limited to certain spots on the road network and small areas, respectively, while national measures like Daytime Running Lights affect the whole of a (driver) population. Therefore, decision-making as well as implementation becomes more complicated as measures leave the local level and advance to the regional and national level. It was agreed that all levels of implementation should be considered during case selection to guarantee an overall analysis of the various decision-making processes. 3. Selecting the cases, preference was given to the measures mentioned in different national road safety programmes. Such programmes are characterized thorough long-term and clearly worked-out methods, as well as a detailed catalogue of measurements. Furthermore, road safety programmes are guaranteed by having passed legislation and having all the necessary financing. By selecting cases already incorporated in road safety programmes, medial as well as political attention for the work of WP4 is at its highest and cooperation of decision-makers is most likely. Besides, consultations with the URG members were carried out, to point to the measures of high interest for different countries. 4. A Cost-Benefit Analysis is sometimes conducted for measures that have already been implemented (ex post evaluation). The goal of such studies is to assess if a certain measure made sense from an economic point of view. However, decision-makers are frequently interested in an ex ante analysis, to compare potential costs and benefits of certain road safety measures that have not yet been implemented. It was agreed that the test cases should present a mixture of both approaches. The selected cases (see Table 1) were carried out by several working groups consisting of one to three WP4 members and a non-specified number of URG members. All relevant steps of applicability testing have been conducted in close cooperation with the user reference group, which gave the users the opportunity to be trained in the application of these tools. 1.2 Evaluation techniques The selected cases should be evaluated using standardized techniques. This section provides a concise description of the main steps and data components, which are needed to perform a Cost-Benefit Analysis (CBA)/ Cost-Effectiveness Analysis (CEA) of a road safety measure2. The description includes: basic formulae, safety effects, implementation units, target accidents, accident costs and implementation costs. The evaluation of WP4 case-studies was performed in line with these evaluation techniques.

2 This is a concise compilation of Chapters 2, 3 of the WP3’s report. More details can be found in the report. Page 8 INTRODUCTION

a. Basic formulae The cost-effectiveness of a road safety measure is defined as the number of accidents prevented per unit cost of implementing the measure: Cost-effectiveness = Number of accidents prevented by a given measure/ Unit costs of implementation of measure For this calculation, the following information items are needed: • A definition of suitable units of implementation for the measure, • An estimate of the effectiveness of the safety measure in terms of the number of accidents it can be expected to prevent per unit implemented of the measure, • An estimate of the costs of implementing one unit of the measure. The accidents that are affected by a safety measure are referred to as target accidents. In order to estimate the number of accidents it can be expected to prevent (or prevented) per unit implemented of a safety measure, it is necessary to: • Identify target accidents, • Estimate the number of target accidents expected to occur per year for a typical unit of implementation, • Estimate the safety effect of the measure on target accidents.

The numerator of the cost-effectiveness ratio is estimated as follows: Number of accidents prevented (or expected to be prevented) by a measure = The number of accidents expected to occur per year X The safety effect of the measure

The benefit cost ratio is defined as: Benefit-cost ratio = Present value of all benefits/ Present value of implementation costs When a CBA is applied, then, besides the above CEA’s components, the monetary values of the measure’s benefits are also required. The monetary values imply, first of all, accident costs and, depending on the range of other effects considered, may also include costs of travel time, vehicle operating costs, costs of air pollution, costs of traffic noise, etc. In order to make the costs and benefits comparable, a conversion of the values to a certain time reference is required. Such an action needs a definition of the economic frame, i.e. the duration of effect (length of service life of the project) and the interest rate, which are those commonly used for the performance of economic evaluations in the country. In a basic case, where the benefits come from the accidents saved only (and no influences on travel expenses and the environment are expected), the numerator of the benefit-cost ratio will be estimated as: Present value of benefits = Number of accidents prevented by the measure X Average accident cost X The accumulated discount factor, where the accumulated discount factor depends on the interest rate and the length of life of the measure.

Page 9 INTRODUCTION

b. Safety effects The most common form of a safety effect is the percentage of accident reduction following the treatment. The main source of evidence on safety effects is from observational before- after studies. Other (theoretical) methods for quantifying safety effects are also possible. One should remember that the safety effect of a measure is stated as available if the estimates of both the average value and the confidence interval of the effect are known. One should also ascertain that both the type of measure and the type of sites (units) for which the estimates are available, correspond to those for which the CBA/CEA is performed. For WP4’s evaluations, it was desirable to apply the local values of safety effects, i.e. those attained by the evaluation studies performed in the country. When the local values do not exist, the summaries of international experience can be used3. If the value of a safety effect is supposed to be provided by a current study (for which the CBA is performed), the estimation of safety effect should satisfy the criteria of correct safety evaluation. This implies that the evaluation should account for the selection bias and for the uncontrolled environment (e.g. changes in traffic volumes, general accident trends). c. Implementation units In the case of infrastructure measures, the appropriate unit will often be one junction or one kilometre of road. In the case of area-wide or more general measures, a suitable unit may be a typical area or a certain category of roads. In the case of vehicle safety measures, one vehicle will often be a suitable unit of implementation, or, in the case of legislation introducing a certain safety measure on vehicles, the percentage of vehicles equipped with this safety feature or complying with the requirement. For police enforcement, it may be a kilometre of road with a certain level of enforcement activity (e.g. the number of man-hours per kilometre of road per year); in the case of public information campaigns - the group of road users, which is supposed to be influenced by the campaign. d. Target accidents The accidents affected by a safety measure present a target accident group. Depending on the type of safety measure it can also be a target injury group, target driver population, etc. Target accidents depend on the nature of the safety measure considered. There are no strict rules for this case. For general measures like black-spot treatment, traffic calming, speed limits, etc. the target accident group usually includes all injury accidents. One should remember that if we apply a specific and not general accident group, proper corrections should be performed for the accident costs, as well.

3 Such as: Elvik R. and Vaa T (2004) The handbook of road safety measures. Elsevier. Page 10 INTRODUCTION e. Accident costs As known, a detailed survey of practice in estimating road accident costs in the EU and other countries was made by an international group of experts as part of the COST- research programme4. Five major cost items of accident costs were identified as follows: (1) Medical costs (2) Costs of lost productive capacity (lost output) (3) Valuation of lost quality of life (loss of welfare due to accidents) (4) Costs of property damage (5) Administrative costs

The relative shares of these five elements differ between fatalities and the various degrees of injuries, and also differ among countries. We assume that each country has its official valuations of accident injuries and damage. Otherwise, the comparative figures from the recent studies can be of help5. All the values are applicable for the WP4’s evaluations but, in every case, there should be a clear indication which components of the above accident costs are included. For the sake of comparability of the evaluation results, the monetary values will be converted to € at 2002-prices. The literature discusses mostly the valuations of fatalities and injuries whereas a CBA usually needs average accident costs. In a simple case, the average accident cost can be estimated as the sum of injury costs multiplied by the average number of injuries with different severity levels, which were observed in the target accidents’ group; the damage value per accident should be stated and added to the injury costs. f. Implementation costs The implementation costs should be determined for each safety measure considered. The implementation costs are the social costs of all means of production (labour and capital) that are employed to implement the measure. The implementation costs are generally estimated on an individual basis for each investment project. As no strict rules are available on the issue, performing a WP4’s evaluation, all the components of the implementation costs should be explained. Typical costs of engineering measures, which are recommended for the CBA evaluations in the country, are desirable. The implementation costs should be converted to their present values, which include both investment costs and the annual costs of operation and maintenance. Similar to the case of accidents costs, for the sake of comparability of the evaluation results, the monetary values will be converted to € at 2002-prices.

4 Alfaro, J-L.; Chapuis, M.; Fabre, F. (Eds): COST 313. Socioeconomic cost of road accidents. Report EUR 15464 EN. Brussels, Commission of the European Communities, 1994. 5 see Chapter 2 of WP3’s Handbook

Page 11 case A: Anti-Lock braking systems for motorcycles

ROSEBUD WP4 - CASE A REPORT

ANTI-LOCK BRAKING SYSTEMS FOR MOTORCYCLES

BY MARTIN WINKELBAUER,

AUSTRIAN ROAD SAFETY BOARD (KFV), AUSTRIA ANTILOCK BRAKING SYSTEMS FOR MOTORCYCLES

TABLE OF CONTENTS

1 PROBLEM ...... 15 2 DESCRIPTION...... 15 3 TARGET GROUP ...... 16 4 ASSESSMENT METHOD...... 16 4.1 Choice of efficiency assessment method ...... 16 4.2 Assessment Tool, Calculation Method ...... 16 4.3 Types of assessed impacts: safety, environment, mobility, travel time ...... 16 4.4 Considered cost of the measure ...... 17 5 ASSESSMENT QUANTIFICATION...... 17 5.1 Target group...... 17 5.2 Accident statistics, number of licensed vehicles...... 18 5.3 Unit of implementation ...... 19 5.4 Crash costs ...... 19 5.5 Vehicle lifespan ...... 20 5.6 "NoVA": the tax to reduce...... 20 5.7 ABS market prices ...... 21 6 ASSESSMENT RESULTS...... 21 7 DECISION-MAKING PROCESS...... 22 8 IMPLEMENTATION BARRIERS ...... 22 9 CONCLUSION/DISCUSSION...... 22 1 PROBLEM ...... 27 2 DESCRIPTION OF THE MEASURE...... 27 2.1 System description...... 28 2.2 Target accident group ...... 29 2.3 Objectives of the measure ...... 29 2.4 Impact of Section Control on average speed ...... 30 3 COST-BENEFIT ANALYSIS...... 31 3.1 Costs of the measure ...... 31 3.2 Economic benefits due to reduced road traffic emissions ...... 31 3.3 Effect on accidents...... 34 3.4 Revenues due to speed violation ...... 38 3.5 Computation of the Cost-Benefit Ratio...... 39 4 CONCLUSIONS...... 40 5 DECISION-MAKING PROCESS...... 41

Page 13 ANTILOCK BRAKING SYSTEMS FOR MOTORCYCLES

CASE OVERVIEW

Measure

1. Fitting motorcycles with anti-lock brake systems (ABS) 2. Reducing vehicle-specific taxes on ABS for motorcycles Problem On the one hand, ABS is highly beneficial in reducing motorcycle accident numbers and severity. On the other hand, ABS is relatively expensive and still not very popular among motorcycle riders, mostly due to the high costs. From the traffic safety point of view, measures must be taken to support ABS equipment for motorcycles, i.e. to raise consumers' willingness to invest in ABS. Target Group Motorcycle riders Targets Reduction of motorcycle accident numbers and severity Initiator Motorcycle dealer organisation Decision-makers Motorcycle dealer organisation, specific motorcycle manufacturer, Ministry of Finance Costs 1. Costs of fitting motorcycles with ABS 2. Tax reduction on this share of the total motorcycle price Benefits Reduction of motorcycle accident numbers and severity, and all related costs. No impacts on the environment, mobility needs and time consumption. Cost-Benefit Ratio

crash reduction potential 8% (min) 10% (max) Cost/Benefit Ratio of ABS 1.11 1.39 Cost/Benefit Ratio of ABS – tax reduction 9.39 11.73

Page 14 ANTILOCK BRAKING SYSTEMS FOR MOTORCYCLES

1 Problem

Elaborate scientific studies clearly indicate that anti-lock brake systems (ABS) are highly beneficial in reducing the number of motorcycle accidents and their severity. But still, only a small number of motorcycle manufacturers offer motorcycles with ABS. Particularly in the cheaper segments, motorcycles with ABS can hardly be found. Only in the segment of expensive motorcycles are ABS frequently offered. It is quite obvious that the price of a heavy, expensive motorcycle covers more easily the cost of fitting it with ABS. Furthermore, only in the expensive segment is ABS frequently found as standard equipment, while in the cheaper segments ABS has to be ordered and paid for separately. It was found that the reasons for motorcycle drivers not to buy a motorcycle with ABS are: • ABS not available in the class of motorcycle they want to buy • ABS not available for the model they want to buy • lack of knowledge on the safety potential • price • biased opinions against ABS If safety features for powered vehicles have to be promoted, tax reductions frequently are named as an effective option. This option has been used effectively several times particularly for measures reducing air pollution from passenger cars.

2 Description

Anti-lock brake systems are a very effective countermeasure against driver misbehaviour in emergency situations. The daily training of a driver - including each and every braking manoeuvre performed - creates a clear message: the closer the stopping distance, the harder you have to brake. In an emergency situation where the expected stopping distance exceeds the available space, the driver takes countermeasures within fractions of a second according to this message. This means that the driver will pull the emergency brake lever as hard as he or she can. This emergency reaction (reflex) cannot be influenced by an average driver and can only be corrected afterwards by experienced and well-trained drivers. For the motorcycle, the reflex of emergency braking usually leads to blocking one or both wheels, which immediately creates a very high danger of falling off the vehicle. Motorcycle drivers are well aware of this danger and leave a huge "safety gap" between the decelerations they actually apply and the real decelerating potential of their vehicles. Motorcycle drivers use practically only about 60% of the decelerating potential of their vehicles [VAVRYN, WINKELBAUER, 1998]. Anti-lock brake systems use different technical approaches. In general what they do is avoid the blocking of wheels during braking. In most of the cases this will keep motorcycle drivers from falling off their vehicles when braking under emergency conditions. In addition, this will also enable motorcycle drivers to significantly improve their braking performance [VAVRYN, WINKELBAUER, 2002] Within this study, two different approaches are assessed. The first approach is anti-lock brake systems itself as a vehicle-based safety measure. The second is tax reduction on

Page 15 ANTILOCK BRAKING SYSTEMS FOR MOTORCYCLES safety features, which generally reduces the price of a "safe vehicle". Applied to this case, a lower price of ABS may encourage motorcycle drivers to buy motorcycles with ABS.

3 Target group

Basically, the target group for this measure is all motorcyclists purchasing a new motorcycle. For the quantification of the safety effect [KRAMLICH, SPORNER, 2000] relevant accident types (target accidents) were identified and the impact was assessed. These results were combined in the end to give figures about the reduction potential on the bases of all motorcycle accidents.

4 Assessment method

4.1 Choice of efficiency assessment method

• ABS was assumed to have an impact on accidents of all severity categories. • Environmental impacts were not expected. • Time consumption impacts were not expected. • Effects on mobility needs were not expected. Although there are only safety impacts to consider as benefits, these effects occur at different levels of severity, i.e. fatal, severe and minor injuries and property damage. None of the categories can be left out due to the size of impact. To combine all of these into a common criterion, a cost/benefit analysis is needed.

4.2 Assessment Tool, Calculation Method

A self-made calculation method was chosen using a spreadsheet program.

4.3 Types of assessed impacts: safety, environment, mobility, travel time

Safety To estimate the direct accident-reducing impact of ABS, a very elaborate study from Germany was used as a reference for this efficiency assessment. Other direct impacts than these were not expected. But it was not obvious how having an ABS on the vehicle changes driver behaviour. Many studies have been performed to assess the safety impact of ABS in passenger cars, most of them detecting that the safety effect of ABS is close to zero. A survey based on accident data from the United States [FARMER et al, 1996] indicates a small but significant increase of fatalities to occupants of ABS-equipped passenger cars. Particularly, fatal single-vehicle crashes are more frequent if cars are fitted with ABS. However, this particular study does not address impacts on other than fatal injuries, and all these studies were based on passenger car accident data. Although single-vehicle accidents are more frequent among motorcycle accidents, the results found for passenger cars cannot simply be adopted for motorcycle accidents. Particularly because the main effect of motorcycle ABS (avoidance of drivers falling of the vehicle instantly after blocking one or both wheels) is not applicable to passenger cars.

Page 16 ANTILOCK BRAKING SYSTEMS FOR MOTORCYCLES

There was no evidence that ABS have any impacts on motorcycle drivers' risk-taking behaviour. Environment, mobility, travel time As indicated above, impacts on the environment, mobility and travel time were not expected.

4.4 Considered cost of the measure

Initially, this study was intended to assess the effectiveness of reducing taxes on ABS, i.e. the share of the total motorcycle price in terms of fitting it with ABS. This is an easy task if ABS is offered as extra equipment and is not included in the regular price. For reasons of comparability and to avoid complexity (not referred to in the studies estimating the accident reduction potential), it was decided to also use these values for motorcycles with ABS as standard equipment.

5 Assessment quantification

5.1 Target group

KRAMLICH and SPORNER published a study on accident reduction potential of motorcycle ABS at the 2000 Ifz motorcycle conference. They identified accident types where ABS may influence accident numbers and severity by using in-depth data from 910 motorcycle accidents that occurred on German roads. Among 610 crashes involving one motorcycle and one passenger car, 65% involved the motorcycle driver using the brake prior to the collision. Among these, 19% of the motorcycle drivers fell off the vehicle. In 93% of these cases ABS would have avoided the crash, or at least reduced the severity of the accident. 300 single-vehicle crashes were identified. 82.7% were accidents at corners (with 40% of the drivers falling off the vehicle before a collision with an obstacle or running off the road) and 17.3% on straight roads (50% drivers falling off). For least 40% of the single-vehicle crashes, ABS would be beneficial by avoiding the accident or at least reducing its severity. Applying these results to all motorcycle accidents including all types of crashes, ABS would be beneficial in 54% of the cases. This gives a final estimate of reducing all fatal and severe injuries to motorcycle drivers by 8 to 10% in Germany. To apply these findings to Austria, two issues had to be checked: • Distributions of accident types in Germany and Austria were compared and were found to be very similar. • There is no evidence that the reduction potential found for each of the accident types should differ between Germany an Austria (e.g. the number of drivers braking prior to the collision). Another question concerned which categories of motorcycles to integrate into the study. The options were: • Light motorcycles: this term changed in definition during recent years; currently this means motorcycles with a maximum of 25 kW engine power and mass/power ratio of

Page 17 ANTILOCK BRAKING SYSTEMS FOR MOTORCYCLES

at least 0.16 kg/kW. This vehicle category has existed since 1991 with the introduction of the graduated licensing system, which defines this category as a novice driver bike. • "Kleinmotorrad": motorcycles with 50 cm³ capacity at most, but no speed limit. This category no longer exists, but there are still several thousand vehicles registered. • Moped: 50 cm³, 45 km/h speed limit. • Motorcycles with a side car. • Motorcycle: more or less all vehicles besides the categories mentioned above. Since the light motorcycle is a sub-category of motorcycle, there is no difference in speed limits and there are similar conditions in daily use. It was decided to select both these categories, i.e. motorcycle and light motorcycle, and to leave out all other categories. Besides, it is very unlikely that mopeds fitted with ABS will be on the market soon (or will have a considerable market share). Driving dynamics of all vehicles running on more than two wheels cannot be compared to powered two-wheelers (PTW).

5.2 Accident statistics, number of licensed vehicles

During recent years, motorcycle accident numbers changed significantly in Austria. The number of licensed vehicles increased enormously. Although the total number of fatalities and injuries has been relatively constant over the last decade, there was a significant shift within the age distribution. While the number of younger accident victims went down, the number of 35 to 55 year old persons injured or killed as motorcycle drivers grew significantly. Particularly due to the strong increasing numbers of registered vehicles, it was decided to focus on recent years. Between 2001 and 2002 the method of collecting data on registered vehicles changed significantly, making data up to 2001 not comparable to later numbers of registrations. Taking all this into account, accident and registration data from 1999 to 2001 was taken as a basis for this assessment. The average of these years was used for calculating total crash costs and crash costs per registered motorcycle. By selecting this method, the latest available accident data without the shortcoming of unsuitable registration data was chosen.

Page 18 ANTILOCK BRAKING SYSTEMS FOR MOTORCYCLES

Table 2: Accident and vehicle statistics, Austria, 1987 - 2001 Motorcycle occupants and registrations in Austria

slight severe fatalities total number of registered injuries injuries vehicles by the end of the year 1987 1718 1,492 104 87,920 1988 1713 1,597 117 99,445 1989 1763 1,524 123 104,840 1990 1664 1,468 96 105,177 1991 1664 1,483 106 112,219 1992 1758 1,452 80 124,904 1993 1519 1,300 96 138,034 1994 1743 1,426 94 154,297 1995 1502 1,256 85 174,907 1996 1470 1,233 84 193,685 1997 1550 1,364 111 212,791 1998 1673 1,446 87 236,314 1999 1833 1,602 103 261,744 2000 1997 1,656 112 278,118 2001 1935 1,628 107 293,053 Mean 99-01 1,921.7 1,628.7 107.33 277,638

5.3 Unit of implementation

There were two options do define a unit of implementation: • The entire vehicle park (i.e. all registered motorcycles in Austria) • one motorcycle To make estimates for the whole vehicle park, it would have been necessary to predict sales statistics on motorcycles in total and the share of motorcycles equipped with ABS. The only advantage would have been to be able to predict the total budget needs when tax reduction is given to safety equipment. Selecting one motorcycle gives a clearer picture of the cost/benefit relation and is independent from future market development.

5.4 Crash costs

The accident costs for Austria were taken from the Austrian Road Safety Programme 2002-2010. The study by METELKA, CERWENKA and RIEBESMEIER (published 1997, data from 1993) used does not include humanitarian costs and added value of the market. As it was agreed upon for all ROSEBUD WP4 case studies, these values were adopted to the 2002 price level. A study to recalculate the accident costs for Austria is currently being prepared and will be supported by the Federal Ministry of Transportation, Innovation and Technology. Referring to the fact that this assessment deals with motorcycle accidents, is was decided to assume the occurrence of major material damage in case of fatal and severe injuries, and minor material damage only in cases of slight injuries.

Page 19 ANTILOCK BRAKING SYSTEMS FOR MOTORCYCLES

Table 3: Crash costs in Austria

1993 2002 fatalities € 805,33 € 949,897 severe injuries € 43,605 € 51,439 slight injuries € 3,695 € 4,359 major material damage € 4,870 € 5,745 minor material damage € 1,242 € 1,465

5.5 Vehicle lifespan

The average lifespan of a motorcycle is a crucial question since it directly impacts the annual implementation costs. Unfortunately this question is very difficult to answer; only some basic conditions could be found to finally reach an estimate. The current vehicle licensing statistics (including all vehicles currently having a licence plate) showed an average age of 8.77 years for motorcycles, 8.06 years for light motorcycles and 8.53 years all together if "old-timers" (first registration 1979 and earlier) are excluded. If these vehicles are included and an average age of 28 years is estimated, the total average age is 11.19 years (9.86 for motorcycles and 13.27 years for light motorcycles). But these numbers include all vehicles currently registered and therefore only determine a minimum for the average lifespan. It was estimated that a motorcycle with one calendar year, on average, is used for 78% of the year, considering the sales per month and the duration of the motorcycle season in Austria from April to October. Some detailed data provided by Honda Austria showed that in the segment of touring and Enduro motorcycles, about 15 years after some representative models were taken from the market, more than 50% of the vehicles once sold were still registered. Crosschecks have been performed by looking at the sales of spare parts that are regularly replaced. This showed that these vehicles are not only in the licensing statistics, but also being used. For the super sport segment, this procedure leads to much shorter estimates for lifespan, what may be caused by the way these vehicles are used and who is using them. In the luxury segment, after 20 years more than 90% of the vehicles are still on the roads. To determine exactly the impact of vehicle park development on accident statistics considering the market penetration with ABS equipped vehicles, detailed data on mileage by vehicle age would have been necessary. Unfortunately such data was not available. Considering all this input, the average lifespan of a motorcycle was estimated to be 12 years.

5.6 "NoVA": the tax to reduce

In Austria, 20% VAT has to be paid in most cases for powered vehicles. Additionally there is the "Normverbrauchsabgabe" ("NoVA"), which can be translated as "fuel consumption tax". For motorcycles, this tax equals 0.02% of the net price multiplied by the capacity in cubic centimetres, then reduced by 100. On average, about 10% NoVA has to be paid (data provided by Honda Austria). Generally, the NoVA percentage is applied to the price of the vehicle including all extras and VAT. It was intended to discount the value of ABS from the NoVA assessment base.

Page 20 ANTILOCK BRAKING SYSTEMS FOR MOTORCYCLES

5.7 ABS market prices

There are different types of ABS systems on the market available at different costs. There is a difference in the construction of these systems, but no evidence of different accident reduction potential. Generally speaking, the cheaper system is used for motorcycles in a lower price segment of vehicles. Without VAT and NoVA the current market prices for the two systems were € 454.55 and € 862.07. Considering a market share of 67% for the cheaper system and transferring to 2002 prices, the average net market price for an ABS was considered to be € 561.11. The average tax reduction would therefore be € 66.39 per vehicle.

6 Assessment Results

The calculation procedure: • Injuries of all severity levels were investigated, evaluated and the numbers from the years 1999 to 2001 were determined to be most useful for further assessment. • Total annual crash costs were calculated in reference to the unit of implementation, i.e. one motorcycle using average numbers of registered vehicles within this period. • Using the minimum and maximum of crash reduction potential, minimum and maximum monetary values for annual cost reductions were calculated. • The average lifespan of a motorcycle was investigated. Using statistics on currently registered vehicles, monthly sales statistics and statistics on specific vehicles comparing sales and number of vehicles still running, the lifespan was estimated at 12 years. • Total cost reduction over the lifespan of a motorcycle was calculated. • ABS market prices were investigated and brought to 2002 price level. • Average tax rates were investigated. • Using all this data, the cost/benefit ratio was calculated for motorcycle ABS and for a NoVA tax elimination on motorcycle ABS. Table 4: costs and benefits of motorcycle ABS over the lifespan of an average vehicle, Austria crash reduction potential costs per vehicle 8% (min) 10% (max) average crash costs € 623.24 € 779.06 ABS costs € 561.11 € 561.11 Cost/Benefit Ratio of ABS 1.11 1.39 Cost/Benefit Ratio of ABS – tax reduction 9.39 11.73

Page 21 ANTILOCK BRAKING SYSTEMS FOR MOTORCYCLES

7 Decision-Making Process

The case study "motorcycle ABS" was intended to be a "real life” case in the frame of ROSEBUD WP4. This included taking the risk of not having feedback from the decision- making process within the duration of WP4. Before starting this case, it was intended to bring the case forward to the Ministry of Finance together with another case, a change of taxes on particle emission filters for diesel engines. This could not be achieved, besides, when this was done, the CBA on motorcycle ABS was not finished. Another attempt to bring the case forward to the Ministry of Finance was not successful. This was the current status when the work on WP4 cases had to be finished. The initial intention to integrate the Austrian Motorcycle Importers' Association into the decision-making process had to be dropped due to political reasons. However, there will be another attempt to reduce the tax on ABS for motorcycles using this CBA as a core argument. If this step should be carried out with the duration of ROSEBUD, the experiences will be considered for the final product and published in the ROSEBUD newsletter.

8 Implementation barriers

Before starting this assessment, frame conditions were scanned for possible barriers, considering the barriers identified in ROSEBUD WP2 and WP3. None of the fundamental barriers seemed to play a significant role within this work. A fundamental question was raised within this study: Is it appropriate to consider tax reductions on safety equipment of vehicles as a road safety measure? This would mean to only take this tax reduction into account as costs of the measure. Or will the entire cost of the safety equipment have to be considered in a public economic sense? Some shortcomings were found in the data available. There is no appropriate data on vehicle mileage, particularly mileage data referring to the age of the vehicle. At the beginning of this study it was clear that tax reductions on safety equipment had never been granted before. Even internationally, no such cases could be found although tax reductions are frequently proposed to promote safety equipment of vehicles.

9 Conclusion/Discussion

General It was proposed by a research institute to carry out this assessment to support motorcycle manufacturers and dealers in their intention to ask the Ministry of Finance for a tax reduction on ABS for motorcycles. Particularly, if the Ministry of Finance is the recipient of such a claim, the cost/benefit assessment seemed to be promising as an argument. • It was unclear whether tax reduction on vehicle safety equipment may exclusively be considered as a cost in the context of public economy, or the entire cost for this safety equipment has to be accounted for.

Page 22 ANTILOCK BRAKING SYSTEMS FOR MOTORCYCLES

• If elimination of the "NoVA" tax on ABS purchase costs in Austria is considered as a road safety measure, it is cost effective by a factor of 9.39 to 11.73 (reduction of fatal and severe injuries by 8 to 10%). • The cost/benefit ratio of fitting motorcycles with an ABS is between 1.11 and 1.39 (reduction of fatal and severe injuries by 8 to 10%) in Austria. Technical • Accident data was easily accessible at an appropriate level of quality. • Vehicle registration data was easily accessible, however, a strong trend during the recent years gave some limitations on the time period to include. • The average lifespan of a vehicle (i.e. motorcycle) could not be determined directly. • There was no appropriate data on vehicle mileage and mileage by vehicle age to exactly determine exposure. • There was good evidence of the impact of the measure. Since this data was from abroad, it was necessary to check its validity in Austria, which was also easy to do. • It was easy to determine the costs of the measure, i.e. ABS market prices and average tax rates. • The calculations could easily be carried out using a spreadsheet program.

REFERENCES

KRAMLICH T., SPORNER, A. (2000): Zusammenspiel aktiver und passiver Sicherheit bei Motorradkollisionen. GDV, Institut für Fahrzeugsicherheit, München. VAVRYN K., WINKELBAUER M. (1996):Bremsverzögerungswerte und Reaktionszeiten bei Motorradfahrern, KfV. Wien. SPORNER A. (1996): Ansatzpunkte für die Bewertung der Risikoexponierung bei PKW/Motorrad - Kollisionen, Büro für Kfz-Technik, VdS. München. VAVRYN K., WINKELBAUER M. (1998): Bremskraftregeverhalten von Motorradfahrern, KfV. Wien. VAVRYN K., WINKELBAUER M. (2003): Bremsbedienung von Motorradfahrern mit und ohne ABS, KfV. Wien. ROSEBUD WP3 Report (2004): Improvements in efficiency assessment tools. ROSEBUD WP2 Report (2004): Barriers to the use of efficiency assessment tools in road safety policy.

Page 23 CASE B1: Section Control – Automatic Speed Enforcement in the Kaisermühlen Tunnel (Vienna, A22 motorway)

ROSEBUD WP4 - CASE B REPORT

SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL (VIENNA, A22 MOTORWAY)

BY CHRISTIAN STEFAN

AUSTIAN ROAD SAFETY BOARD (KFV), AUSTRIA SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

TABLE OF CONTENTS

1 PROBLEM ...... 27 2 DESCRIPTION OF THE MEASURE...... 27 2.1 System description...... 28 2.2 Target accident group ...... 29 2.3 Objectives of the measure ...... 29 2.4 Impact of Section Control on average speed ...... 30 3 COST-BENEFIT ANALYSIS...... 31 3.1 Costs of the measure ...... 31 3.2 Economic benefits due to reduced road traffic emissions ...... 31 3.3 Effect on accidents...... 34 3.4 Revenues due to speed violation ...... 38 3.5 Computation of the Cost-Benefit Ratio...... 39 4 CONCLUSIONS...... 40 5 DECISION MAKING PROCESS...... 41

Page 25 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

CASE OVERVIEW

Measure Section Control - Automatic Speed Enforcement in the Kaisermühlen Tunnel (Vienna, A22 motorway)

Problem Traffic accidents due to excessive speeding Target Accident Group All accidents in the tunnel Objectives Reducing accidents and harmonization of traffic flow (reduction of “Stop-and-Go” traffic or congestion during peak hours) Initiator Austrian highway operator (ASFINAG) Decision makers Austrian highway operator (ASFINAG), Federal Ministry of Transport, Innovation and Technology, Federal Ministry of the Interior, local government of the municipality of Vienna Costs Capital costs are divided into costs for construction and maintenance costs; investments into the construction of the Section Control are covered by the ASFINAG, whereas operating costs are covered by the Federal Ministry of the Interior Benefits Benefits include reductions in accidents and savings in road traffic emissions. Running costs of the system are cleared by fines from speed violators Cost-Benefit Ratio Cost-Benefit Ratio for tunnels on urban motorways: 5.4

Page 26 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

1 Problem

Exceeding the speed limit is probably the most common law violation among drivers. Yet, only a small proportion of all traffic violators are detected, i.e. the risk of being apprehended is usually very low. According to the Federal Ministry of the Interior, inappropriate speed is responsible for more than a third of all fatal accidents occurring on Austrian roads. Measures to reduce the percentage of speeders would therefore amount in a significant reduction of both casualty accidents and severity of injuries. Speed limits are usually set in accordance with road conditions, traffic volume, proximity to sensitive areas, such as residential areas and schools, and a host of other factors. Motorists are expected to obey posted speed limits at all times. Traditional manual and stationary speed enforcement methods are limited in their effects and require a lot of human resources. Automatic speed enforcement on the other hand is intended to provide enhanced capacity for enforcement by applying technical solutions that do not require the presence of police officers at the scene of an offence. Systems for automatic speed enforcement (including Section Control) are designed to detect and identify traffic violators automatically. Identification is solely based on photographs of the vehicle or the driver, respectively.

2 Description of the measure

The Kaisermühlen Tunnel is an urban tunnel with separate tubes for each direction of traffic. More than 90,000 vehicles use this part of the A22 motorway everyday; about 10% consist of Heavy Goods Vehicles (HGV). Due to a nearby tank lot, the share of HGV carrying flammable liquids (e.g. motor spirits, diesel oil) is extremely high. The tunnel offers 3-4 lanes per direction with entrance and exit ramps within the tunnel.

Figure 1: Site overview of the Section Control in the Kaisermühlen Tunnel

Page 27 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

Table 5: Road characteristics of the Kaisermühlen Tunnel

KAISERMÜHLEN TUNNEL Road classification Urban motorway (A22) Type of road Tunnel with two tubes Number of lanes per direction 3-4 Width per lane 3.5 m Length 2.3 km Passenger cars, buses, motorcycles: 80 Speed limit km/h Heavy Goods Vehicles (>7,5 t): 60 km/h Daily traffic (20036) 91,915 vehicles/24 hours Amount of Heavy Goods Vehicles (HGV) 10.0% Source: Vienna Municipal Department 34, own calculations

2.1 System description

In close cooperation with the Federal Ministry of Transport, Innovation and Technology, the Federal Ministry of the Interior and the municipality of Vienna, the Austrian highway operator (ASFINAG) introduced a new instrument of traffic surveillance to reduce accidents and traffic delays in the Kaisermühlen Tunnel on one of Vienna’s most frequented motorways (A22) in August 2003. This so-called Section Control does not measure speed at a certain point in space and time, but calculates the average speed by means of passage time in a defined area (see Figure 2). The aim is to force drivers not only to slow down at certain points of stationary speed control (e.g. automatic speed cameras), but also adhere to the speed limit over the entire distance. It also provides live monitoring of traffic flow behaviour and thus contributes to harmonizing traffic flow performance.

The system consists of two facilities, one for each driving direction. Vehicle detection is carried out optically. A video system placed above the road on gantries (one camera above each of the three lanes) takes two pictures of each passing vehicle, one at the beginning of the tunnel and one at the end. These photographs provide details of the event (passage time, use of lane) and the license plate number. Furthermore a laser scanner installed adjacent to the video system is programmed to differentiate between passenger cars and lorries (HGV), which is fundamental to keep different speed limits under surveillance.

At the entrance and exit of the Kaisermühlen Tunnel, laser scanners are installed to obtain the required data. The system continually looks for two matching license plates - if a match is found, the average speed is calculated and if it exceeds a defined level, an image of the license plate is transmitted to the traffic supervision department. This information is used to establish the owner of the vehicle via the national motor vehicle and driver’s license registration database. Data of vehicles not exceeding the pre-set speed limit (plus a

6 Computed data by means of a linear regression model. Vehicle data related from the automatic counting station have been inadequate due to false HGV readings in one direction.

Page 28 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL certain tolerance) are deleted immediately afterward. Only aggregated data is kept for statistical reasons. Figure 2: Scheme of Section Control in the Kaisermühlen Tunnel

Source: Vienna Municipal Department 34

The Section Control system is designed to operate with speeds up to 250 km/h and a maximum traffic flow of 2 vehicles per second and lane. Vehicle detection is independent of the position of a vehicle on or between lanes. There is no necessity for pavement installations (like inductive loops) or disruption of the traffic flow.

2.2 Target accident group

The target accident group of this measure consists of accidents occurring in the Kaisermühlen Tunnel. This survey concentrates on injury accidents because data for material damage accidents could not be collected without enormous strains on budget and working hours. Thus, the cost-benefit ratio computed in the following chapters underestimates the real impacts on accidents to a certain extent. This should be kept in mind whenever Section Control systems are considered for further use in traffic safety programmes.

2.3 Objectives of the measure

The main task of Section Control is the measurement of average speed of motor vehicles for the purpose of speed control and traffic enforcement. Contrary to the majority of commonly used speed control systems, which mostly operate in combination with Doppler radars, the Section Control system supervises the traffic performance along a defined road section. It also offers a wide range of additional features regarding traffic surveillance. Objectives • Monitoring different speed limits that apply to different vehicle classes • Harmonization of traffic flow (reduction of “Stop-and-Go” traffic or congestion during peak hours) • Surveillance of closed lanes (in combination with route information and management systems)

Page 29 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

• Detection of wrong-way drivers (“ghost cars”) • Image triggering (including alarm release) for vehicles exceeding height limits • Detection of stolen vehicles • Traffic surveillance (for the tunnel operator) • Statistical data (traffic speed, loads, headways)

2.4 Impact of Section Control on average speed

According to the Federal Ministry of the Interior7, in 2003 more than 35% of fatal accidents on roads in Austria occurred because of inappropriate speed. As mentioned in the previous chapter, the main objective of Section Control is harmonization of speed, which has a positive influence on accidents. In its first year of operation, a reduction in average speed by more than 10 km/h was recorded (see Figure 3). Traditional mobile and stationary speed surveillance (in use before the Section Control started operating) showed the average speed of all vehicles to be 85 km/h, whereas this value decreased to about 70 km/h shortly after the introduction of the measure. Further speed measurements carried out after a 6-month period revealed that average speed on this road section has levelled off to 75 km/h due to the fact that drivers tend to follow regulations in a very strict manner right after their implementation, but less some time afterwards due to unintended behavioural adaptations (”kangaroo effect“). Drivers started acting in accordance with the speed limit as soon as technical installations were established, and reports about this new system of speed control appeared in the media. Figure 3: Effect of Section Control on average vehicle speed

Source: Vienna Municipal Department 34

In close cooperation with local police services and employees of the Institute for Driver Education and Vehicle Technology of the Austria Road Safety Board (KfV), the following distinction in average speed of passenger cars and HGV during daytime (5 am - 10 pm) and night time (10 pm - 5 am) was made. This breakdown is essential for calculating detailed traffic emissions and fuel consumption for different road users.

7 KfV, 2004, page 50

Page 30 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

Table 6: Average speed of passenger cars and lorries before and after implementation of Section Control

Passenger cars HGV

Before After Before After Daytime 85 km/h 75 km/h 70 km/h 55 km/h Night time 95 km/h 75 km/h 75 km/h 55 km/h Source: own estimates in cooperation with local police services

3 Cost-Benefit Analysis

3.1 Costs of the measure

Investment costs for the Section Control in the Kaisermühlen Tunnel add up to € 1,200,000 (2003 price). Construction work of gantries, cables and data lines to the Section Control server are included in this price. Annual costs of operation and maintenance are about € 60,000, covering a service contract of 4 service cycles per year plus additional repairs if the system starts malfunctioning. In order to avoid disruption of traffic flow, maintenance and repairs are done during night hours when traffic is usually very low. According to the Austrian highway operator (ASFINAG), the Section Control system has a 10-year service life, beginning in 2003. After that period, software problems and missing spare parts for the hardware are expected to affect full operation of the system. Investment costs are incorporated in the form of an annual capital cost assuming a 4 percent interest rate in real terms (see Table 7). For the sake of comparability, all costs were converted to their 2002-price level. Total annual costs for operating the Section Control add up to € 204,272 per year. Table 7: Total annual costs of Section Control in the Kaisermühlen Tunnel

EURO EURO (2003-price) (2002-price)

Total annual Expense factors Costs Costs Annual capital costs [n=10, 4% p.a.] costs

Investment costs 1,200,000 1,178,782 145,333 204,272 Annual maintenance costs 60,000 58,939 Source: Vienna Municipal Department 34, own calculations

3.2 Economic benefits due to reduced road traffic emissions

Road traffic is a major source of air pollution and emission of greenhouse gases in Austria. Although improvements in vehicle technology, the introduction of exhaust treatment systems (catalytic converters), and the development of higher quality fuels have to some extent significantly reduced emissions from vehicles, this effect has levelled off by a still ongoing increase in traffic performance. According to latest studies8, traffic volume in and around Vienna will rise by more than 90% by 2035 due to a steady increase in resident population, decentralization and daily distances covered.

8 SAMMER et al, 2004, page 25

Page 31 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

As stated in the previous chapter, a major effect of Section Control is harmonization of velocity, i.e. vehicle drivers maintain a constant speed, reducing “Stop-and-Go” traffic and congestion. The model9 used for computing the resulting changes in road traffic emissions was created by the Austrian Umweltbundesamt, the governmental authority for protection and control of the environment, in close cooperation with associated institutes in Germany and Switzerland. The “Handbook of Emission Factors for Road Transport” provides emission factors in g/km for all current vehicle types (passenger cars, Light Duty Vehicles, Heavy Goods Vehicles and motorcycles), each divided into different categories for a variety of traffic situations. The following parameters have been used to define the model:

• Type of emission: hot emissions, cold start emissions, evaporation • Vehicle type: passenger car - Heavy Goods Vehicle (HGV) • Estimated changes in composition of the vehicle fleet (2003-2013)

• Air pollutants (CO, NOx, SO2, PM10, VOC) and carbon dioxide (CO2) • Type of road: urban motorway • Time of day: daytime/night time

Table 8 gives values for both air pollutants and CO2 as the most important greenhouse gas emitted by road traffic. As can be seen from the annotations in the footnote, different literature sources were used to obtain monetary estimations for the most important air pollutants emitted during combustion. To arrive at 2002 prices, German Mark (DM) and Norwegian Krona (NOK) were first converted into Austrian Shillings (ATS) and then brought to a 2002 price level by using official inflation rates (see appendix). Values of traffic emissions were finally converted to € by multiplication with 0.07267. Table 8: Valuation of environmental impacts for use in cost-benefit analyses

Value per unit Air pollution Unit of valuation DM (1995)10 NOK (1995)11 € (2002)

Tons of NOx- CO 12 1700 974.64 Equivalent

NOx kg of NOx 115 14.90

SO2 kg of SO2 37 4.79

Particle (PM10) kg of PM10 1800 233.27 VOC kg of VOC 15 1.94

CO2 Tons of CO2 220 28.51 Source: own calculations

For quite some years, considerable efforts have been made by the European Commission to reduce fuel consumption and, consequently, emissions of carbon dioxide. In 1992, the Auto-Oil I Program was introduced within the European Union to define emission ceilings

9 KELLER, HAUSBERGER, 2004 10 EWS, 1997, page 41 11 ELVIK, 1999, page 24 12 Conversion factor: 1 ton of CO = 0.003 tons of NOx-Equivalent (EWS, 1997, page 41)

Page 32 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

(EURO classes) for passenger cars as well as Heavy Goods Vehicles, and to set quality standards for fuels for 2000 and beyond.

One key measure in this respect was a voluntary agreement with car manufactures to reduce CO2 emissions from new passenger cars to 140 g/km by the year 2008/2009. For the Kaisermühlen Tunnel, this boost in vehicle technology, along with a lower average speed due to Section Control, results in more than 12,000 tons of saved CO2 emissions, having a discounted monetary value of more than € 280,000 (see Table 9).

Table 9: Monetary value of saved emissions due to Section Control (accumulated value 2003-2013)

Changes in road Discounted value of traffic traffic emissions (t) emissions in € (2002-price) CO - 14.9 -137

NOx - 39.0 -431,639

SO2 - 0.4 -1,552 Particle (PM10) - 0.5 -87,029

VOC + 7.3 +11,247

CO2 - 12,879.6 -281,973 Accumulated value -791.084

Monetary value of saved emissions per year -79,108 Source: Austrian Umweltbundesamt, own calculations

Nitrogen oxide emissions are among the most harmful of all air pollutants. Thus, various nitrogen oxide catalytic converters have been developed which will help to reduce emissions of NOx significantly over the next 10 years. Expected changes can be seen in Table 9, which states above all a constant decrease in saved nitrogen oxide emissions because of improvements in vehicle technology. In the year 2003 nearly 6 tons of NOx were saved through Section Control. This value decreases to one ton of NOx in 2013. Calculated over the economic lifetime of the Section Control system, savings in NOx emissions amount to a value of more than € 430,000.

Volatile organic compounds (VOC), in combination with nitrogen oxides, are responsible for ground level ozone and smog. VOC are primarily produced when fuels are incompletely combusted. Looking at the VOC traffic emissions in the period under observation, an increase of one ton in 2003 and slightly less in the following years has been calculated. This is due to the fact that most vehicle engines have their lowest VOC output between 80 and 100 km/h. A decrease in average speed to 75 km/h (passenger cars) or 55 km/h (HGV) amounts to an increase of VOC emissions.

Page 33 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

Figure 4: Changes in the emission of air pollutants due to Section Control

VOC CO NOx PM10 SO2

1

0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

-1

-2

-3

Changes in road traffic emissions [t] emissions traffic road in Changes -4

-5

-6 Period under observation

Source: own calculations

3.3 Effect on accidents

In its first year of operation, a positive impact of Section Control concerning accidents in the Kaisermühlen Tunnel was observed. Apart from the reduction in total numbers of casualty accidents, the severity of injury was also positively affected. In a four-year period prior to the start of the Section Control system (Ib-IVb), one fatality, one person severely and 10 slightly injured have been recorded on average every year. Since August 2003 no fatal or severely injured road user was observed in the Kaisermühlen Tunnel, while the number of slightly injured drivers decreased to a total of 7 in the after-period (see Table 10). Table 10: Injury accidents before and after the implementation of Section Control Injury Seriously Slightly From To Period Fatalities accidents injured injured

12.08.1999 12.08.2000 IVb 7 1 0 10

12.08.2000 12.08.2001 IIIb 7 0 1 9

12.08.2001 12.08.2002 IIb 7 1 1 11

12.08.2002 12.08.2003 Ib 7 0 0 9

12.08.2003 12.08.2004 Ia 5 0 0 7

Mean (IVb – Ib) 7.0 0.5 0.5 9.8 Source: own calculations

Accidents are statistically rare events. Part of the nature of such events is that the precise time and place of their occurrence, as well as the precise nature of their impacts, are hardly predictable, i.e. in some periods the recorded number of accidents on given points of the road network are greater (or less) than the average values expected for those points. In Figure 5, the grey dots represent the recorded number of accidents and slightly

Page 34 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL injured road users in the Kaisermühlen Tunnel (fatal and serious injuries were omitted due to small numbers). The white dots show the moving average of the annual counts. In the first year, this is the same as the number of accidents or slightly injured for that year. In the second year, it is the average of the first two years, in the third year, it is the average of the first three years, etc. It can be seen that the recorded number of slightly injured road users in a given year is not necessarily representative of the mean annual number. The annual recorded number of slightly injured, for example, varies between 9 and 11. Thus, if a safety inspection leads to choosing these points for treatment, a selection bias occurs and, in the measurements made after the treatment, an effect of diminution is registered (regression to the mean) independent of the treatment. The average value of the four years prior to the installation of Section Control (Ib-IVb) have been chosen as the base for a medium-long term trend.

Figure 5: Recorded number of accidents and slightly injured in the Kaisermühlen Tunnel – mean of the annual numbers

Recorded number of accidents Annual mean accidents Recorded number of slightly injured Annual mean of slightly injured

12

10,0 11 9,8 10 9,5 10 10,0

9 9 8 7777

7,0 7,0 7,0 7,0 6

4 Number of accidents/slightly injured

2

0 IVb IIIb IIb Ib Before periods Source: own calculations To properly quantify the safety effect of Section Control, a simple before/after comparison of accidents is not suitable. It is necessary to compare the situation with Section Control (“after”) with the anticipated situation that would have occurred without Section Control. The latter presents a calculated value of a previously observed (“before”) situation. Therefore, various types of risk indicators (fatality rate, rate of severely injured road users, etc.) and their means and standard deviations were computed (see Table 11).

Traffic performance in the before period (Ib-IVb) increased in a linear manner, while in the after-period (Ia) a slight drop in vehicle-km was observed. This phenomenon is due to the fact that traffic capacity on this road section has apparently reached its limit. Without further investments in additional lanes or route information and management systems, a further increase in daily traffic is unlikely. Because numbers of fatal and serious injuries are too low to produce meaningful results, these two categories were combined for further calculations. Furthermore, some effects of serious injuries on the quality of life (e.g. lifelong paraplegia) deem it necessary to ascribe these victims the same weight as fatalities.

Page 35 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

Table 11:Traffic performance and accident rates [per million vehicle-km] in the Kaisermühlen Tunnel

Traffic Rate of fatal performance Rate of slight Period and serious [million vehicle- Accident rate injuries injuries km]

IVb 67.6 0.10 0.015 0.15

IIIb 70.3 0.10 0,014 0.13

IIb 72.2 0.10 0,028 0.15

Ib 74.8 0.09 0,000 0.12

Ia 74.5 0.07 0,000 0.09

Mean (IVb - Ib) 0.10 0.014 0.14

Standard deviation (IVb - Ib) 0.004 0.011 0.015 Source: own calculations The corrected “before” value (number of accidents, fatalities or injured people without treatment) results from multiplying the average number of accidents (per million vehicle- km) in Table 11 with the traffic performance in the “after” period (Ia). The ratio of “after” and (corrected) “before” values constitutes the actual safety effect of the measure.

Table 12: Corrected before and after values of accident severity due to Section Control

Corrected before value After value Ratio13

Injury accidents 7 5 0.71 Fatal and serious 1 0 0.00 injuries Slightly injured 10 7 0.70

Source: own calculations The analysis also controls for general trends in the number of accidents by using the total number of accidents on motorways in the “before” and “after” period as a comparison group (see Table 13). The mean number of comparison group accidents in the before period was 2,485, respectively, and 2,540 in the “after” period. Thus, the number of comparison group accidents is sufficiently large to be only minimally influenced by random fluctuations. The effect of Section Control on the number of accidents (or fatalities or injured road users) was estimated as follows:

Safety effect [%] = 1- [Xa/E(m)b] / [Ca/Cb] whereas

Xa = recorded number of accidents in the “after” period E(m)b = expected number of accidents (correct before value) in the “before” period Ca = number of comparison group accidents in the “after” period Cb = number of comparison group accidents in the “before” period

13 Slightly different numbers due to round off errors in the computation of the ratio

Page 36 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

Table 13: Injury accidents and severity of casualties on Austrian motorways in the before/after period Injury Seriously Slightly From To Period Fatalities accidents injured injured

12.08.1999 12.08.2000 IVb 2,535 134 1,218 2,847

12.08.2000 12.08.2001 IIIb 2,468 165 1,255 2,703

12.08.2001 12.08.2002 IIb 2,402 121 1,173 2,663

12.08.2002 12.08.2003 Ib 2,534 124 1,133 2,819

12.08.2003 12.08.2004 Ia 2,440 108 1,165 2,642

Mean (IVb – Ib) 2,485 136 1,195 2,758 Source: Road Accident Database of the Austrian Road Safety Board (KfV)

Statistical inference draws conclusions about a population based on sample data. It also provides a statement, expressed in the language of probability, of how much confidence we can place in the conclusions. The different values for the safety effect of Table 14 acts as estimators of the (unknown) population parameter. The purpose of a confidence interval is to estimate this parameter with an indication of how accurate the estimate is and how confident we are that the result is correct. Any confidence interval consists of two parts: an interval computed from the data and a confidence level. The confidence level states the probability that the method will give a correct answer. That is, if you use a 95% confidence interval, the probability that the true value is out of this interval is only 0.05.

Table 14 and Table 15 show estimates and 95% confidence intervals of the safety effects of Section Control on accidents. Computing the Odds Ratio, note that if any value out of 4 numbers involved in the evaluation is zero, a correction must be applied, i.e. 0.5 should be added to each number.14 Table 14: Safety effect of Section Control on accident severity

Odds ratio Safety effect [%]

Injury accidents 0.69 -30.5

Fatal and serious injuries 0.34 -66.4

Slightly injured 0.72 -28.4 Source: own calculations

Table 15: Best estimate and confidence interval of the safety effect of Section Control on accidents

Percentage change in the number of accidents

Accident severity Best estimate 95% confidence interval Injury accidents -31 (-35; -26) Fatal and serious -66 (-82; +143) injuries Slightly injured -28 (-39; -13) Source: own calculations

14 FLEISS, 1981, page 64

Page 37 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

Table 16 gives an economic valuation of savings in the number of accidents and severity of injury due to Section Control. The original values were obtained from a study on economic costs of accidents15. Figures were then converted into EURO (€) and brought to a 2002 price level by using official inflation rates (see appendix). As can be seen from the bottom line of the table, the safety effect of the Section Control system amounts to annual savings of more than 1 million €.

Table 16: Valuation of savings in the number of accidents and severity of injury due to Section Control

Amount of € per unit Category Cumulated value savings (2002-price)

Fatalities 1 949,897 949,897 Seriously 1 51,439 51,439 injured Slightly 3 4,359 13,077 injured Property 2 5,745 11,490 damage

Total 1,025,903 Source: own calculations

3.4 Revenues due to speed violation In the period under observation (13.09.2003 - 27.08.2004), more than 29 million vehicles passed through the Kaisermühlen Tunnel and about 40,000 drivers were charged because of excessive speeding (see Table 17). That is, only 0.14% or every 700th driver, does not follow speed regulations on this road section and drives too fast. The top speed of a vehicle heading north was 175 km/h and 154 km/h heading south. About 5% (2,161) of all fines issued were acquired by HGVs. Keeping in mind that more than 10% of daily traffic is due to HGVs, a possible explanation for this phenomenon can be found in the high proportion of foreign vehicles among lorries. Due to the fact that mutual recognition of financial penalties has only been established with Germany and Switzerland, most of the foreign speed violators cannot be prosecuted.

Table 17: Speed violations and charges in the Kaisermühlen Tunnel

Vehicles passing Fines the Section All Passenger cars HGV Control vehicles Heading 13,450,345 19,162 951 20,113 south (A23) Heading north 15,973,473 19,558 1,210 20,768 (Stockerau) Total 29,423,818 38,720 2,161 40,881 Source: Federal Ministry of the Interior, own calculations

15 BUNDESMINISTERIUM FÜR WISSENSCHAFT UND VERKEHR, 1997, page 136-141

Page 38 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

At the Tampere European Council (15 and 16 October 1999), the Heads of State or Government of the EU-Member states and the President of the Commission agreed that mutual recognition of criminal and financial matters should be a cornerstone of judicial cooperation within the European Union. Thus, France, the United Kingdom and Sweden initiated the adoption of a Council Framework Decision that enables member states to execute criminal and financial offences against citizens of other member states. Although this proposal is far from reaching legal status due to objections from several countries, it can be expected to pass legislation within the next 3-5 years. Obtaining fines from foreign speed violators should then be possible and benefits will be maximized.

According to Austrian law16 80% of the fines from speed violations belong to the operator of the infrastructure, which (in case of the Section Control) is the Austrian highway operator (ASFINAG). The remaining 20% are used to cover the maintenance costs of the system settled by the Federal Ministry of the Interior.

Table 18 gives fines for different levels of speeding. Drivers exceeding the speed limit by more than 50 km/h have their driving licences revoked. During the observation period, this happened in 46 cases. Table 18: Revenues due to excessive speeding in the Kaisermühlen Tunnel

Revenues due to Fine Violators speed violation 0 – 9 km/h € 21 16,176 339,696

10 – 19 km/h € 42 22,048 926,016

20 – 29 km/h € 56 2083 116,648

30 – 39 km/h € 70 409 28,630

40 – 50 km/h € 140 119 16,660

Total 40,881 1,427,650 Source: Federal Ministry of the Interior, own calculations

3.5 Computation of the Cost-Benefit Ratio

The Cost-Benefit Analysis is based on the principle of economic efficiency, i.e. to estimate if a measure is worth being implemented, the benefits and costs of the treatment are computed and brought into relationship. The benefit term includes all positive (monetary) effects of the measure. In the case of Section Control, benefits consist of reductions in accidents and road traffic emissions. Revenues from speed violators were omitted in the calculation of the Cost-Benefit Ratio because of the fact that in an economic point of view, it is irrelevant if the money belongs to consumers buying goods and therefore increasing their personal benefits or the highway operator that uses the fines for additional safety campaigns. The Cost-Benefit Ratio will be the same at both events. Different benefits are added to obtain a total benefit. The cost term on the other hand denotes implementation and maintenance costs.

16 StVO, Article 100, Paragraph No.10

Page 39 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

The Cost/Benefit-Ratio (CBR) is defined as:

Present value of all benefits CBR = Present value of implementation costs

Combining the benefits and costs calculated in the previous chapters, a net present value of all benefits (without fines from speeders) of € 1,105,011 and costs of € 204,272 is obtained (see Table 19). Both values amount to a Cost/Benefit-Ratio of 5.4. According to analyses of safety measures in Work Package 1 of ROSEBUD17, measures with a CBR larger than 3 are ranked “excellent”. Table 19: Present value of benefits and costs in € (2002-price) due to Section Control Components of the CBA Benefits Costs

Road traffic emissions 79,108

Accident costs 1,025,903 Installation and maintenance 204,272 costs Total 1,105,011 204,272

Source: Austrian Umweltbundesamt, Federal Ministry of the Interior, Vienna Municipal Department 34, own calculations

4 Conclusions

The results of the Cost-Benefit Analysis lead to the following conclusions: • Although accidents rates in the Kaisermühlen Tunnel were already well below average (0.12 injury accidents per million vehicle-km on Austrian motorways), a positive safety effect of Section Control was achieved. It can be estimated that the effect would be even more convincing if this safety measure had been implemented to road sections with accident rates above the average. In the weeks to come, another Section Control system will start operating on the motorway A2 near mount “Wechsel”. Previous studies showed that this road section has an accident rate three times above the average. Thus, an even better safety performance than the Section Control in the Kaisermühlen Tunnel can be expected. • This survey concentrates on injury accidents because data for material damage accidents could not be collected without enormous strains on budget and working hours. Thus, the Cost-Benefit Ratio computed underestimates the real effects to a certain extent. This should be kept in mind whenever Section Control systems are considered for further use in traffic safety programs. • Due to the fact that mutual recognition of financial penalties only exists with Germany and Switzerland, most of the foreign speed violators cannot be prosecuted. As soon as the Council Framework Decision on mutual recognition of

17 Road Safety and Environmental Benefit-Cost and Cost-Effectiveness Analysis for Use in Decision-making. ROSEBUD is a thematic network funded by the European Commission to support users of efficiency assessment tools at all levels of government.

Page 40 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

criminal and financial matters has reached legal status, obtaining fines from foreign speed violators should be possible and benefits will be maximized. • With the instrument of Cost/Benefit Analysis, it is possible to incorporate various effects of this safety measure into the evaluation process, i.e. not only reductions in casualty accidents and severity of injuries, but also impacts on the environment, such as road traffic emissions. A major problem of road traffic, which has been neglected due to the special situation of the Kaisermühlen Tunnel, is traffic noise. Regional governments in Austria have already expressed their intention to use Section Control as a means to reduce traffic noise in residential areas. Such an application of Section Control will raise the Cost-Benefit Ratio even more. • The effects of Section Control are closely related to outside influences such as annual average daily traffic (AADT), accident rates, amount of HGVs, etc. That is, if you change the site you will probably get different results than the ones present in this case study.

5 Decision-Making Process

The results of Cost-Benefit Analysis (CBA) on Section Control were presented to officials of the Austrian highway operator (ASFINAG) to answer the question whether this method will be taken into consideration in the future. Regarding the use of Efficiency Assessment Tools (EAT) such as CBA in the decision making process, it was stated that at the time being, such instruments were too complex. Candidates for the introduction of further Section Control systems on the existing road network will initially be detected by comparing accident and fatality rates of road sections with the motorway average of this type of road. The decision whether or not Section Control is an appropriate instrument to reduce accident risk is then made after thorough analysis of cause and type of accidents on this specific section. Further concerns were expressed that the results and methodology of EAT are hard to communicate to the public. The more complex the decision making process, the more likely it would be that people mistrust those findings. Another aspect regarding the use of EAT is politically motivated. In the aftermath of catastrophic accidents, such as the fire in the Tauern Tunnel (1999), political pressure concerning a second tube became so high that even if a CBA had led to a negative Cost-Benefit Ratio, this measure would have been implemented nonetheless. Although it is unlikely that CBA will be used in decision making in the near future, officials of the ASFINAG considered Efficiency Assessment Tools an adequate instrument in those cases where decisions cannot be made solely based on accident statistics. Changes in ASFINAG policy could also lead to an increased demand of Efficiency Assessment Tools in the decision making process. As soon as environmental aspects, such as traffic emissions and traffic noise, are considered as important as improving traffic safety, instruments including those aspects are to become an essential part in decision- making. Till then the most influencing factors are accidents and fatality rates and the amount of daily traffic, respectively.

Page 41 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

References

[1] AUSTRIAN FEDERAL ECONOMIC CHAMBER (WKO), Inflation rates in Austria in the years 1996-2002, http://wko.at/statistik/prognose/inflation.pdf, Date of inquiry: 29.09.2004 [2] AUSTRIA ROAD SAFETY BOARD (KfV): “Road Traffic Accidents in Austria”, In: Verkehr in Österreich, Edition No. 36. Vienna, 2004 [3] BUNDESMINISTERIUM FÜR VERKEHR, WISSENSCHAFT UND VERKEHR: “Österreichische Unfallkosten- und Verkehrssicherheitsrechnung Straße“, In: Forschungsarbeiten aus dem Verkehrswesen, Band 79. Wien, 1997 [4] ELVIK, R.: “Cost-benefit analysis of safety measures for vulnerable and inexperienced road users”, Work package 5 of EU-Project PROMISING, TØI- Report 435, Institute of Transport Economics. Oslo, 1999 [5] EUROPEAN UNION (EU): „Screening of efficiency assessment experiences“, Report “State of the Art”, Work package 1 of EU-Project ROSEBUD. July 2003 [6] FLEISS, J.: „Statistical methods for rates and proportions“. New York, 1981 [7] FORSCHUNGSGESELLSCHAFT FÜR STRASSEN- UND VERKEHRSWESEN, Arbeitsgruppe Verkehrsplanung: “Empfehlungen für Wirtschaftlichkeitsunter- suchungen an Straßen (EWS) - Entwurf, Aktualisierung der RAS-W 86. 1997 [8] KELLER, M.; HAUSBERGER, S.; et al: „Handbuch der Emissionsfaktoren des Straßenverkehrs in Österreich“, Version 2.1 erstellt im Auftrag von Umwelt- bundesamt, Ministerium für Land- und Forstwirtschaft, Umwelt und Wasser- wirtschaft sowie dem Bundesministerium für Verkehr, Innovation und Technologie. Vienna, 2004 [9] OANDA.COM – The currency site, FXHistory: historical currency exchange rates, http://www.oanda.com/convert/fxhistory, Date of inquiry: 26.07.2004 [10] ROAD ACCIDENT DATABASE of the Austrian Road Safety Board (KfV), Date of inquiry: 18.10.2004 [11] SAMMER, G.; ROIDER, O.; KLEMENTSCHITZ, R.: “Mobilitäts-Szenarien 2035 - Initiativen zur nachhaltigen Verkehrsentwicklung im Raum Wien”, Editor: Shell Austria GmbH. Vienna, 2004 [12] STRASSENVERKEHRSORDNUNG (StVO) 1960, Article 100, Paragraph No. 10, Website of the Austrian Federal Chancellery: http://www.ris.bka.gv.at, Date of inquiry: 19.10.2004 [13] VIENNA MUNICIPAL DEPARTMENT 34 - Building and Facility Management, City Administration of Vienna

Page 42 SECTION CONTROL – AUTOMATIC SPEED ENFORCEMENT IN THE KAISERMÜHLEN TUNNEL

Appendix

Table A1: Inflation rates in Austria in the years 1996-2002 Year Inflation [%] 1996 1.9 1997 1.3 1998 0.9 1999 0.6 2000 2.3 2001 2.7 2002 1.8 Source: http://wko.at/statistik/prognose/inflation.pdf

Table A2: Average currency exchange rates for different European countries

Exchange rate From To Period (annual mean) DM ATS 1995 7.04001 NOK ATS 1995 1.59133 ATS EURO 2002 0.07267 Source: http://www.oanda.com/convert/fxhistory

Page 43 CASE B2: Automatic Speed enforcement on the A13 motorway (NL)

ROSEBUD WP4 - CASE B REPORT

AUTOMATIC SPEED ENFORCEMENT ON THE A13 MOTORWAY (NL)

BY CHRISTIAN STEFAN

AUSTIAN ROAD SAFETY BOARD (KFV), AUSTRIA AUTOMATIC SPEED ENFORCEMENT ON THE A13 MOTORWAY (NL)

TABLE OF CONTENTS

1 INTRODUCTION...... 47 2 DESCRIPTION OF THE MEASURE...... 47 2.1 System description...... 48 2.2 Objectives ...... 49 2.3 Improving traffic safety ...... 49 2.4 Harmonisation of traffic flow...... 50 2.5 Reduction of air pollution...... 50 2.6 Reduction of traffic noise...... 51 3 COST-BENEFIT ANALYSIS...... 51 4 CONCLUSIONS...... 51

Page 45 AUTOMATIC SPEED ENFORCEMENT ON THE A13 MOTORWAY (NL)

CASE OVERVIEW

Measure Automatic Speed Enforcement (Section Control) on the A13 motorway in Overschie

Problem Traffic accidents, noise and air pollution due to excessive speeding Target Accident Group All accidents on the A13 motorways Objectives Reducing accidents and harmonization of traffic flow (reduction of traffic emissions and traffic noise due to lower speed limit) Initiator National Police Service Agency KLPD Decision-makers National Police Service Agency KLPD, Ministry of Transport Costs No data available Benefits Benefits are reductions in accidents, greenhouse gas emissions and traffic noise Cost-Benefit Ratio Could not be calculated due to missing data

Page 46 AUTOMATIC SPEED ENFORCEMENT ON THE A13 MOTORWAY (NL)

1 Introduction

In the Netherlands, speed enforcement is the most important task of the motorway police. Since the implementation of general speed limits and the beginning of structural speed control in May 1988 (by the National Police Service Agency KLPD), motorists have been acting at a level of major violation. In December 1993, a pilot for Continuous Applied Speed Enforcement (CASE1) was started by the KLPD and the Ministry of Transport on the A2 between Utrecht and Amsterdam. Before the implementation of the pilot, the speed limit was violated by 35% of the motorists, increasing to almost 70% during the night. After CASE1 started operating, speed violations decreased to almost 3%. This result led to an institutionalization of Automatic Speed Enforcement in 1995, becoming a part of daily operational procedure.

2 Description of the measure

In May 2002, the Dutch authorities introduced a Section Control system on the motorway A13 aimed at maintaining the maximum speed limit at 80 km/h. One of the main purposes of this measure was to improve the air quality in Overschie, a municipality of Rotterdam. About 124,000 vehicles use this motorway everyday, which includes almost 10% of Heavy Goods Vehicles (HGV). As the A13 crosses through a densely populated area, noise and air pollution have become a major cause of distress for local residents.

Another objective of the Section Control system was to reduce the number of accidents and severity of injury, respectively. The National Traffic Safety Policy of the Netherlands aims at reducing the number of fatalities by 50% and injuries by 40% (in comparison to 1985) by the year 2010.

Figure 6: Site overview of Section Control on the A13 motorway

Source: MALENSTEIN, 2003, page 3

Page 47 AUTOMATIC SPEED ENFORCEMENT ON THE A13 MOTORWAY (NL)

Table 20: Road characteristics of the A13 motorway

A13 MOTORWAY Road classification Urban motorway Number of lanes per direction 3 Width per lane 3.5 m Length 2 km Speed limit 80 km/h (all vehicles) Daily traffic 124.000 vehicles/24 hours Amount of Heavy Goods Vehicles (HGV) 10% Source: MALENSTEIN, 2003, page 10; TNO, 2003, page 5

2.1 System description

A Section Control system was set up over a 2 km stretch on the A13 in Overschie. A video system placed on gantries on both sides of the control zone captures and stores an image of each passing vehicle. These images are reduced to a limited amount of information, providing a digital fingerprint for every vehicle. The Section Control server continually searches for two matching fingerprints. If a match is found, the computer calculates the average speed and stores both images as one object on a permanent medium if this value is above a pre-set margin. A nearly invisible flashlight on the gantries allows the system to function during low light conditions without blinding the drivers.

Recognition of the license plate is handled by a separate application. Checking the vehicles’ categories (passenger car, lorry, motorcycle, etc.) is done via a special license database in the Ministry of Transport. The length of each passing vehicle is measured by inductive loops.

Before the Section Control was set in force, it had to be guaranteed that fines could not be appealed in court. Thus, a police patrol of the Traffic and Transport Division deliberately committed a speed offence, which was registered by the Section Control system. This offence was taken to a Dutch court as a test trial. During the process, technical details and the mode of operation of Section Control were explained and accepted as evidence by the judges. When the patrolman was convicted, he appealed and matters were taken to the next higher court. When he was convicted again, legislation was achieved.

Page 48 AUTOMATIC SPEED ENFORCEMENT ON THE A13 MOTORWAY (NL)

Table 21: Statistics of the Section Control system on the A13 motorway

A13 MOTORWAY Accuracy of measurement < 1% error Accuracy of vehicle identification 99.75% Accuracy of license plate recognition 84.8% Detection of speed up to 250 km/h certified (156 mph) Start of Section Control 11.05.2002 Fully automated (15.6% of the violators have Processing of violators to be processed manually due to errors in license plate recognition)

Before implementation of Section Control ⇒ 6,000 violators/day (4.8% of daily traffic) Violations After implementation of Section Control ⇒ 700-800 violators/working day (0.6%) ⇒ 1,000-1,100 violators/weekend (0.9%) Source: MALENSTEIN, 2003, page 17

Questionnaires among motorists showed a surprisingly high rate of acceptance of Section Control. 75% of the interviewees considered this system to be more reasonable than traditional speed enforcement (radar traps). Combined with sufficient information on the road, the methodology of Section Control was appreciated because of its structured approach. Motorists experienced that there was no escape and obediently followed speed regulations. The major effects of this measure were slowing of traffic and a better use of the infrastructure.

2.2 Objectives

The main task of Section Control is the measurement of average speed of motor vehicles for the purpose of speed control and traffic enforcement. This objective was triggered by the National Traffic Safety Policy to reduce the number of fatalities by 50% by the year 2010. Due to harmonization of traffic flow, Section Control allows for a better use of the existing infrastructure and reductions in traffic emissions and traffic noise. Objectives • Improving road safety • Harmonisation of traffic flow • Reduction of air pollution • Reduction of traffic noise

2.3 Improving traffic safety

Accident data before and after the implementation of Section Control on the A13 was not available. According to Jan Malenstein from the Dutch National Police Agency (KLPD), the safety effect of continuous speed enforcement (implemented in the Netherlands in 1993)

Page 49 AUTOMATIC SPEED ENFORCEMENT ON THE A13 MOTORWAY (NL) accounts for -20% in injury accidents and -25% in the number of fatalities over a period of 10 years.

2.4 Harmonisation of traffic flow

Based on loop detectors at the beginning and the end of the control zone, average speed and traffic flow before and after the implementation of Section Control was monitored. Analysis of the measurements showed a clear decrease in average speed and v85 after Section Control started operating - speed fluctuations became smaller and extreme peaks occurred less often. Speed measurements carried out after several months revealed a slight increase in average speed. Traffic behaviour was adapted due to continuous speed control, resulting in a harmonized traffic flow (reduction of “Stop-and-Go” traffic) and less congestion. Calculations showed a decrease of congestion during peak hours by 30%.

2.5 Reduction of air pollution

The assessment of air quality before and after the introduction of Section Control was based upon measurements and modelling. An hour-to-hour line-source model was applied to compute the contribution of traffic emissions on the A13 to air quality in Overschie. Continuous monitoring of NO, NO2 and PM10 was performed at three different locations: one 500m west of the A13 (“background location”) and the other two 50m and 200m east of the motorway. Measurements were carried out between April 2001 and April 2003, including periods of one year before and one year after the implementation of Section Control. Furthermore, NO2 concentrations were monitored with passive samplers at more than 30 locations in Overschie between April 2002 and April 2003.

At a limited number of locations, black smoke and concentrations of elemental and organic carbon (ES and OC) were measured. Meteorological data and traffic data on the A13 were obtained from the Meteorological Services KNMI and the Netherlands Road Directorate (RWS). Data from road loops provided information on the number, category and speed of vehicles before and after the measure. In addition, TNO provided emission factors specifically derived for the A13 before and after the implementation of Section Control.

The main findings and conclusions are as follows:

Section Control has been effective in reducing fluctuations in traffic flow and speeding (especially during the night). Traffic moving at a constant, moderate speed emits less air pollutants compared to traffic with high speed fluctuations. Measurements carried out after Section Control started operating on the A13 showed that traffic flowed more efficiently through Overschie, although the number of vehicles has increased drastically in the past years. Compared to a motorway with the same amount of traffic, this measure is estimated to reduce NOx emissions by 15-25% and PM10 by 25-35%, respectively (see Table 22).

Measurements of NO2 concentrations in Overschie with passive samplers indicate that at a distance of 250m from the A13, impacts of traffic emissions were no longer detectable. Model calculations were used to assess the effect of Section Control on air quality. NO2 - concentrations in a distance of 200m east of the motorway decreased by 25% and 34% for PM10, respectively. It has to be emphasized that these results are specific for Overschie. At other locations different ratios of passenger cars and HGV, or different traffic dynamics and congestion conditions, would influence the impacts of continuous speed control in a way that might be quite different from the situation in Overschie. Thus, it is recommended

Page 50 AUTOMATIC SPEED ENFORCEMENT ON THE A13 MOTORWAY (NL) to perform specific research for each location before implementing a Section Control system. Table 22: Changes in the emission of air pollutants on the A13 motorway due to Section Control

Changes in Air pollutants emissions

Traffic emissions NOx -15 – 25% Traffic emissions PM10 -25 – 35% Concentration of air pollutants NO -25% at a distance of 200m 2 Concentration of air pollutants PM10 -34% at a distance of 200m Source: TNO, 2003, page 6

Regarding environmental aspects, continuous speed control is an important instrument to reduce traffic emissions as long as more source-orientated measures (e.g. less polluting vehicles, “clean” fuels, less road traffic) are not available.

2.6 Reduction of traffic noise

In addition to environmental and safety aspects, Section Control also reduced traffic noise by forcing drivers to follow the reduced speed limit of 80 km/h. Research on traffic noise before and after the measure was implemented and showed a significant reduction in the noise level by 5.6 dB(A). However, this result cannot be solely attributed to the reduction in maximum speed, but also to the changing of the top layer of the A13. Local authorities recommended new test trails along a 25m pathway on both carriageways to eliminate this influence.

3 Cost-Benefit Analysis

The Cost-Benefit Ratio (CBR) could not be calculated due to missing data (costs of the measure). Concerning the benefits of Section Control, most of the information (accident data, reduction of greenhouse gases, etc.) was available in aggregated form only. In order to compute monetary values for those benefits, original data would have to be used.

4 Conclusions

The Section Control on the A13 was highly successful in achieving the preset objectives. Speed violations have been reduced, average speed decreased and extreme speed violations have become an exception. Based on model calculations, the reduction of average speed also had a positive impact on traffic emissions and traffic noise. 75% of the motorists approve of Section Control because they experience less traffic congestion during peak hours.

Page 51 AUTOMATIC SPEED ENFORCEMENT ON THE A13 MOTORWAY (NL)

References

[1] MALENSTEIN; J.: Madrid ITS Japan Session Segment Control, Presentation documents of the Section Control System on the A13. Madrid, 2003 [2] MALENSTEIN; J.: VERA2 – The issue of cross border enforcement in the Netherlands, Presentation for the European Commission on speed enforcement. 2003 [3] The Netherlands Organisation for Applied Scientific Research (TNO): “Onderzoek naar effecten van de 80 km/u- maatregel voor de A13 op de luchtkwalitweit in Overschie”, TNO Report 258. Apeldoorn, 2003

Page 52 CASE C1: Daytime Running Lights in The Czech Republic

ROSEBUD WP4 - CASE C REPORT

DAYTIME RUNNING LIGHTS IN THE CZECH REPUBLIC

BY PETR POKORNÝ

TRANSPORT RESEARCH CENTRE, CDV, THE CZECH REPUBLIC DAYTIME RUNNING LIGHTS IN THE CZECH REPUBLIC

TABLE OF CONTENTS

1 PROBLEM TO SOLVE ...... 56 2 DESCRIPTION...... 57 3 TARGET GROUP ...... 57 4 ASSESSMENT METHOD...... 57 5 ASSESSMENT QUANTIFICATION...... 58 6 ASSESSMENT RESULTS...... 61 7 DECISION MAKING PROCESS AND BARRIERS ...... 61 8 CONCLUSION/DISCUSSION...... 62

Page 54 DAYTIME RUNNING LIGHTS IN THE CZECH REPUBLIC

CASE OVERVIEW

Measure Implementation of Daytime Running Lights (DRL) during the entire year Problem The high number of casualties in daytime multi-party accidents (target accident group) Target Group Drivers and owners of motor vehicles Targets Implementation of DRL, which will lead to the significant reduction of casualties in daytime multi-party accidents Initiator The first initiator will be the Transport Research Centre, which will provide the results of this CBA to the Ministry of Transport Decision Makers In case of potential implementation of the measure, the Ministry of Transport will elaborate and incorporate a relevant amendment into the Road Act; the Parliament will then have to authorise it. Costs All costs are calculated for a 12-year period, which is the lifetime of DRL automatic switches. All monetary values are converted to 2002 prices. The following costs were calculated: • the cost of automatic light switches • maintenance and repair costs of these switches • additional replacement costs of bulbs due to wear The total costs are € 70,410,000 for 12 years Benefits Positive benefit • reduction in casualties (48 fatalities are estimated to be prevented due to DRL annually) Negative benefits • extra fuel costs due to DRL • environmental effects The total benefits are calculated to be € 303,570,000 for 12 years Cost-Benefit Ratio 1/4.3

Page 55 DAYTIME RUNNING LIGHTS IN THE CZECH REPUBLIC

1 Problem

The Czech Republic has a high number of casualties caused by the road accidents (compared to most other EU countries). The implementation of DRL would contribute to the decrease of this number. The implementation of DRL will improve the visibility of motor vehicles in daytime, which will lead to a decrease in multi-party daytime accidents. It will also contribute to lower collision speeds in accidents involving DRL-equipped motor vehicles. The vehicles will be more visible; drivers will be able to react faster in the case of a potentially dangerous situation and can start to slow down earlier. This will also have a significant effect on the number of casualties. Figure 7: Comparison of total numbers of road accident fatalities in selected European countries from 1980 – 2003

.

Source: CDV

Table 23: Numbers of casualties (until 30 days after accident), 2002

Fatalities 1,431 Severely injured* 5,492 Slightly injured 29,013 Source: Summary of accidents data, the Traffic Police Directorate of CZ, 2003 *The definition of severely injured is a person who spends a minimum of 7 days in hospital due to a road accident.

Page 56 DAYTIME RUNNING LIGHTS IN THE CZECH REPUBLIC

2 Description

2.1 Definition of DRL

This measure is a legal obligation for all motor vehicles to drive with low beam headlights on or with special DRL lamps during the whole year [ETSC, 2003]. For this calculation the use of special DRL lamps is not considered. For the calculation it is assumed that an automatic light switch is installed in all new vehicles from January 2002 onwards. This means that in all older vehicles, low beam headlights have to be switched on manually or the automatic light switch will be installed additionally. Mopeds and motorcycles are not considered in this calculation because DRL has already been obligatory for them. Another aspect to consider is the current use of DRL, which will have an effect on the calculation. The following calculation assumes the effect of DRL on target accident fatalities to be 20%. The DRL effect on the number of casualties is higher than on the number of multi- party daytime accidents, which can be explained because of lower collision speeds [ETSC, 2003]. The number of fatalities in the target accident group (multi-party daytime accidents) was estimated to be 30% of all fatalities. Because it was not possible to find the relevant number in Czech national statistics, the estimation was made based on Austrian statistics.

2.2 Legal situation

DRL has been obligatory for mopeds and motorcycles throughout the whole year since 1.1.2001. For other motor vehicles, DRL is obligatory in winter (from the last Sunday in October to the last Sunday in March – for this study the winter time lasts 5 months). This obligation is stated in the National Road Act [§ 32, law 361/2000].

3 Target Group

The target group is drivers and owners of motor vehicles.

4 Assessment method

CBA was applied in the calculation because it enables on to evaluate the monetary valuation of the measure’s benefits and costs. CBA provided in 2003 by ETSC [“Cost Effective Transport Safety Measures”] was used as the basis, and was also an important source of information and assumptions. In order to make the costs and benefits comparable, the duration of effect was formulated. The duration of the measure is determined for 12 years – this is the entire lifetime of DRL automatic switches in cars. The effects of DRL are calculated for 7 months in a year. The fact, that a lot of road users use DRL on a voluntary basis is also considered. The estimation that 10% of drivers have already been using DRL was made. It is also assumed, that 90% of drivers will use DRL when it becomes obligatory.

Page 57 DAYTIME RUNNING LIGHTS IN THE CZECH REPUBLIC

For the sake of comparability of the evaluation results, the monetary values are converted to € at 2002 prices. To calculate the present value of benefits and costs, the accumulated discount factor of 5% is assumed. The safety effect of DRL is calculated for fatalities prevented; target accidents are multi- party daytime accidents. The impacts of DRL are as follows: • Safety effects – using DRL will lead to a 20% reduction in the number of fatalities of target accidents. The proportion of injuries and property damage is included in the cost of one fatality prevented. It is estimated that 30% of total fatalities occur in DRL- relevant accidents. • Environmental effects – using DRL will lead to extra fuel consumption. The additional contribution to air pollution due to DRL use for all vehicles is about 1% of the total cost of pollution arising as a result of fuel emissions in road transport [ETSC, 2003]. • Additional fuel costs due to DRL (price of fuel excluding tax and VAT) – for passenger cars this consumption is estimated to be 0.1 l/hour in traffic, while for trucks it is 0.12 l/hour in traffic. Costs considered: • The price of automatic light switches in new cars is estimated at € 5. The price of retrofitting amounts to € 40 including installation costs per vehicle. It is estimated that 10% of old vehicles will install the automatic light switch [own estimation]. • Maintenance and repair costs of automatic light switches during its lifetime are estimated at € 10 [own estimation]. • Additional replacement costs of bulbs related to ‘wear and tear’ of the bulbs during daytime – additional bulb costs are € 2 per car per year [own estimation]. It is assumed that the costs do not affect mobility.

5 Assessment Quantification

5.1 Safety effect

Safety effects are calculated only for reduction of the number of fatalities. The reduction of injuries and property damage is included in the calculation of the cost of one fatality prevented. The cost of one fatality prevented was determined to be € 1,076,000. This amount was calculated based on “Socio-economy losses caused by accidents in CZ in 2002” [KOŇÁREK, 2002]. The cost of one fatality prevented includes medical costs, costs of lost productive capacity (lost output) and administrative costs. The proportional share of the costs of minor and serious injuries and property damage is also considered in the cost of one fatality.

Page 58 DAYTIME RUNNING LIGHTS IN THE CZECH REPUBLIC

It is estimated that 30% of the total fatalities occur in DRL-relevant accidents and that DRL will lead to a 20% reduction in the number of fatalities. The reduction of fatalities is calculated as follows: The number of fatalities * the average 90% use of DRL * the 30% of the DRL-relevant accidents * the 20% effect of DRL for fatalities [ETSC, 2003].

Table 24: Numbers of fatalities (until 30 days after accident), 2002

1.4.2002 – 30.10.2002 Number of fatalities 899 DRL-related fatalities 270 Fatalities prevented by DRL 48 Source: Summary of accidents, the Traffic Police Directorate, 2003 In 12 years, the total cost of fatalities prevented (including proportional costs of injuries and property damage) is € 460,230,000.

5.2 Cost of extra fuel

Due to the large differences in fuel consumption it is not suitable to calculate average fuel consumption. As the extra DRL fuel consumption is independent of the standard fuel consumption of vehicles, the time that a vehicle participates in traffic was calculated. The extra fuel consumption by DRL is 0.1 l/h (0.1 litre of fuel during 1 hour of drive) for passenger cars and 0.12 l/h for trucks. The average distance driven in one hour is estimated at 50 km on all types of roads. The share of km driven during the daytime is 55% of the total sum of vehicle km [ETSC, 2003]. Required data: • Number of vehicles and million vehicle-km The number of passenger cars was 3,650,000 in 2002 and number of trucks was 460,000 in 2002 [Czech Ministry of Interior]. Number of vehicle-km is not known, so the estimation had to be done – on average, a passenger car drives 10,000 km a year and a truck 30,000 km a year [ETSC, 2003]. Table 25: Numbers of cars and mill. Vehicle-km in 7 months, 2002

Passenger cars 3,650,000 Trucks 460,000 Daytime mill. vehicle-km - cars 11,700 Daytime mill. vehicle-km - trucks 4,430 Source: The Czech Statistical Office, own estimation Average 2002 price of fuel excluding tax and VAT. Table 26: Price of fuel excluding tax and VAT, the year 2002

Diesel € 0.315 Petrol € 0.308 Source: CDV

Page 59 DAYTIME RUNNING LIGHTS IN THE CZECH REPUBLIC

5.2.1 The calculation of extra fuel consumption due to DRL

The correction due to the voluntary use of DRL is 20%. A correction is needed because 10% of car users already use DRL on a voluntary basis and 90% will use DRL after the law is set. The correction value is 0.8. Passenger cars - 11,706 mill. Vehicle-km / 50 km * 0.8 * 0.1 l = 18,730,000 litres Trucks – 4,430 mill. Vehicle-km / 50 km * 0.8 * 0.12 l = 8,500,000 litres The fuel costs for cars and trucks are € 8,300,000 in 2002. In 12 years, the total cost is € 73,960,000.

5.2.2 Environmental costs

The cost of pollution arising as a result of DRL extra fuel emission is about 1% of total pollution costs caused by fuel emissions in road transport [ETSC, 2003]. In the Czech Republic, the average estimation price of external costs from road emissions for 2002 is calculated to be € 1,600,000,000 [CDV]. A cost of € 82,700,000 is calculated for the 12- year period due to DRL use.

5.3 Calculation of other costs

5.3.1 Automatic light switch

The price of an automatic light switch in new cars is estimated at € 5. The number of new cars sold in 2002 was 170,000. The price of retrofitting amounts to € 40, including installation costs per old vehicle. It is estimated that 10% of old vehicles will install the automatic light switch [ETSC, 2003, own estimation]. The total costs for 12 years are € 23,700,000.

5.3.2 Maintenance and repair costs of automatic light switches during its lifetime

The costs are estimated at € 10 for one car equipped with a light switch [ETSC, 2003, own estimation]. The total cost for 12 years is € 17,100,000.

5.3.3 Additional costs as a result of the wear of the bulbs during daytime use

The replacement rate for bulbs increases by a factor of 1.4 for the Czech Republic. The additional bulb costs are € 2 [ETSC, 2003, own estimation]. The correction of 0.8 is needed. The total cost for 12 years is € 29,610,000.

Page 60 DAYTIME RUNNING LIGHTS IN THE CZECH REPUBLIC

6 Assessment Results

Table 27: Costs and benefits – 12-year period

Fatalities 460,230,000 € prevented Extra fuel -73,960,000 €

Emission costs -82,700,000 € Total benefits 303,570,000 €

Light switches 23,700,000 €

Maintenance 17,100,000 € Bulbs 29,610,000 € Total costs 70,410,000 € Cost / benefit 1/4.3

7 Decision-Making Process and Barriers

In case of potential implementation of DRL, the measure has to be part of the Road Act and must be ratified by the national parliament. This situation is the main barrier – some decisions of parliament members are not based on rational reasons (e.g. CBA), but on political or personal opinions. Especially in the case of DRL (and other road-related laws), some members of parliament assume themselves to be road experts just because they drive many kilometres per year. The role of CDV in this process is vital – CDV should introduce the results of this CBA (and other related CBAs) to the members of the Subcommittee on Road Safety and to disseminate the results between the experts. Based on a survey amongst decision makers (members of the parliament of the Czech Republic: Ms Soňa Paukrtova - Chairman of the Subcommittee on Road Safety of the Senate of the Czech Parliament, Mr Miroslav Fejfar - also member of this Subcommittee, Ms Ivana Večeřová - Secretary of the Economical Committee of the Senate), the following general conclusions could be drafted:

• CBA could be one of the most important tools to force implementation of road safety legislative measures in relatively short amount of time. • CBA could play a key role in the decision-making process, especially at present when there is a lack of public finance sources in the Czech Republic. • To increase the usage and to widespread CBA amongst decision-makers, wider dissemination of information on CBA is vital. Research institutes should play a more active role in this process. Nevertheless, processes in the parliament are mostly political ones so one could also expect negative reactions, or not taking CBA into account due to the political reasons.

Page 61 DAYTIME RUNNING LIGHTS IN THE CZECH REPUBLIC

8 Conclusion/Discussion

The calculation has shown that the use of DRL would significantly contribute to improving the road safety situation in both countries and that making DRL obligatory would bring significant benefits to the whole society. The difference in the CBA-results in Austria and in the Czech Republic is caused by the lack of some data for the Czech Republic, which therefore had to be estimated. The main barrier in Austria preventing the obligatory use of Daytime Running Lights consisted of objections by stakeholders (e.g. drivers’ unions) in technical and social aspects of this measure. Additional fuel consumption and the fear of elderly drivers getting stranded because of dead batteries have been major arguments in past discussions. Recent developments in vehicle technology (automatic switches for DRL) could dispel most of those objections. In the Czech Republic, the wider discussion regarding making DRL obligatory during whole year has not started yet. There is a common understanding that the current situation (DRL obligatory only in the winter season) is sufficient enough. The introduction of results of several international studies (including this one) to the relevant decision-makers seems to be the first step in the process of constant DRL implementation. Barriers in the implementation process are expected. These barriers could occur from political reasons and also from the technical point of view (like in Austria). Therefore, providing independent and current information regarding benefits of DRL is vital for a potential beginning of the implementation process.

REFERENCES

ETSC (2003): Cost Effective Transport Safety Measures The Czech Statistical Office: www.czso.cz The Czech Ministry of Interior: http://www.mvcr.cz/statistiky/crv.html KOŇÁREK (2002): Socio-economy losses caused by accidents in CZ The Traffic Police Directorate of CZ (2003): Summary of accidents data

Page 62 CASE C2: daytime running lights in AUSTRIA

ROSEBUD WP4 - CASE C REPORT

DAYTIME RUNNING LIGHTS IN AUSTRIA

BY PETR POKORNÝ

TRANSPORT RESEARCH CENTRE, CDV, THE CZECH REPUBLIC DAYTIME RUNNING LIGHTS IN AUSTRIA

TABLE OF CONTENTS

1 PROBLEM TO SOLVE ...... 66 2 DESCRIPTION...... 67 3 TARGET GROUP ...... 67 4 ASSESSMENT METHOD...... 67 5 ASSESSMENT QUANTIFICATION...... 68 6 ASSESSMENT RESULTS...... 71 7 DECISION MAKING PROCESS AND BARRIERS ...... 71

Page 64 DAYTIME RUNNING LIGHTS IN AUSTRIA

CASE OVERVIEW

Measure Implementation of Daytime Running Lights (DRL) during a whole year period Problem High number of casualties in daytime, multi-party accidents (target accident group) Target Group Drivers and owners of motor vehicles Target Implementation of DRL, leading to a significant reduction of casualties in daytime multi- party accidents Initiator Ministry of Transport, Austrian Road Safety Board (KfV) Decision-makers Ministry of Transport Costs All costs are calculated for a 12-year period, which is the usual lifetime of a DRL automatic switch. All monetary values have been converted to 2002 prices. The following costs were calculated: • the cost of automatic light switches • maintenance and repair costs of switches • additional replacement costs of burned out bulbs The total costs are € 195,300,000 for 12 years. Benefits “Positive” benefits • reduction of casualties (53 less fatalities per year are estimated due to DRL in Austria) “Negative” benefits • extra fuel costs due to DRL • environmental effects The total benefits are calculated to be € 695,000,000 for 12 years. Cost-Benefit Ratio 1/3.6

Page 65 DAYTIME RUNNING LIGHTS IN AUSTRIA

1 Problem

In Austria, the number of accidents (especially fatalities) looks more favourable than relevant data in the Czech Republic. However, any further decrease in these numbers is desirable. DRL is a measure that could contribute to a significant reduction of human fatalities in traffic accidents. Figure 8 shows different trends in Austria and the Czech republic from 1989 – 2002. Figure 8: Trends of development in the Czech Republic and Austria from 1989 – 2002

140% 126%

120% Czech Republic

100%

80% Road accident fatalities 60%

32% 37% 40% Population 20% 5% 0% Passenger cars -1,5% -20%

-40% - 39% -60%

Source: CDV and KfV

The following Table 28 shows the number of casualties in the year 2002.

Table 28: Numbers of casualties (fatalities until 30 days after accident) in 2002

Fatalities 956

Severely injured 14,628

Slightly injured 42,056

Source: KfV Definition of severely injured Whether an injury is severe or slight is determined by §84 of the Austrian criminal code [StGB]. A severe injury is one that causes a health problem or occupational disability longer than 24 days, or one that "causes personal difficulty". An injury or health problem that "causes personal difficulty" is one that affects an "important organ", if it results in a "health disadvantage", if the "healing process is uncertain", or if it leads to the fear of "additional effects”.

Page 66 DAYTIME RUNNING LIGHTS IN AUSTRIA

2 Description

2.1 Definition of DRL

This measure is a legal obligation for all motor vehicles to drive with low beam headlights or special DRL lamps during the whole year [ETSC, 2003]. In the following calculations, the use of special DRL lamps has not been considered. It is assumed that an automatic light switch is installed in all new vehicles from January 2002 onwards. This means that in all older vehicles, low beam headlights have to be switched on manually or the automatic light switch will be installed on them additionally. Mopeds and motorcycles are not considered in this calculation, because DRL is already obligatory for those vehicles. Another aspect needing consideration is the current use of DRL, which has an effect on the calculation. The following calculations assume the effect of DRL on the target accident fatalities to be 20%. The DRL effect on the number of casualties is higher than on the number of multiparty daytime accidents, which can be explained due to lower collision speeds [ETSC, 2003].

2.2 Legal situation

Except for mopeds and motorcycles, DRL is not obligatory in Austria. Austrian law [§ 99 KFG] states certain conditions for using driving lights: running lights have to be switched on at dusk, nightfall and during the night, in fog, or when the overall weather conditions deem it necessary.

3 Target Group

Drivers and owners of motor vehicles.

4 Assessment method

CBA was applied in the calculation because it enables the monetary valuation of the measure’s benefits and costs. CBA provided in 2003 by ETSC [Cost Effective Transport Safety Measures] was considered as the main source for information and assumptions. In order to make costs and benefits comparable, a duration of the effect was needed. The duration of the measure is determined for 12 years – which is the lifetime of DRL- automatic switches in passenger cars. The fact that a lot of road users switch on DRL voluntarily has also been considered. In Austria, nationwide surveys from 1999 and 2003 showed that about one third (1999: 29%, 2003: 37%) of all drivers have already been using DRL. Thus, a share of 35% is used in this calculation. It is also assumed, that 90% of drivers would use DRL when it is obligatory. For the sake of comparability of the evaluation results, the monetary values are converted to € at 2002 prices. To calculate the present value of benefits and cost, an accumulated discount factor of 5% is estimated.

Page 67 DAYTIME RUNNING LIGHTS IN AUSTRIA

The safety effect of DRL is calculated for the number of fatalities saved, target accidents are multiparty daytime accidents. The number of target accident group fatalities was 296 (31% of all fatalities) in the year 2002 [KfV]. The impacts of DRL are as follows: • Safety effects – using DRL will lead to a reduction of 20% in the number of target accident fatalities. The proportion of injuries and property damage is included in the cost of one fatality saved. • Environmental effects – using DRL will lead to extra fuel consumption. For passenger cars this consumption is estimated to be 0.1 l/hour in traffic, while for trucks it is 0.12 l/hour in traffic. The additional contribution to environmental pollution due to DRL use for all vehicles is about 1% of the total cost of pollution arising as a result of fuel emissions in road transport [ETSC, 2003]. • Additional fuel costs due to the DRL (price of fuel excluding tax and VAT).

Considered costs: • The price of an automatic light switch in a new car is estimated at € 5. The price of retrofitting amounts to € 50, including installation costs per vehicle. It is estimated that 15% of old vehicles will install the automatic light switch [ETSC, 2003]. • Maintenance and repair costs of automatic light switches during its lifetime are estimated to be € 15 for Austria [ETSC, 2003]. • Additional replacement costs of bulbs related to the “wear and tear” of bulbs during daytime – the additional bulb costs are € 6 per car and year [ETSC, 2003]. It is assumed that the costs do not affect mobility.

5 Assessment Quantification

5.1 Safety effect

The safety effects are calculated only in reduction of number of fatalities. The reduction of injuries and property damage is included in the calculation of the cost of one fatality saved. The cost of one fatality saved was determined to be € 2,200,000. The cost of one fatality saved includes medical costs, costs of lost productive capacity (lost output) and administrative costs. The proportional share of the costs of minor and serious injuries and property damage is also considered in the cost of fatality. DRL will lead to a 20% reduction in the number of fatalities in the target accident group. The reduction of fatalities is calculated as follows: The number of target accident group’s fatalities * the average 90% use of DRL * the 20% effect of DRL on fatalities [ETSC, 2003].

Page 68 DAYTIME RUNNING LIGHTS IN AUSTRIA

Table 29: Numbers of fatalities (until 30 days after accident), the year 2002

Number of fatalities 956 DRL related fatalities 296 Fatalities saved by DRL 53 Source: KfV In 12 years, the total cost of fatalities saved (including proportional cost of injuries and property damage) is € 1,040,000,000.

5.2 Cost of extra fuel

Due to the large difference in fuel consumption it is not suitable to use an average fuel consumption in the following calculations. As the extra DRL fuel consumption is independent on the standard fuel consumption of vehicles, the time that a vehicle participates in traffic was calculated. The extra fuel consumption by DRL is 0.1 l/h (0.1 litre of fuel during one hour of driving) for passenger cars and 0.12 l/h for trucks. The average distance covered in a one-hour drive is estimated at 50 km on all types of roads. The share of km driving during the daytime is 55% from the total sum of vehicle-km [ETSC, 2003]. Required data: • Number of vehicles and million vehicle-km The number of passenger cars was 4,000,000 in 2002 and number of trucks was 330,000 in 2002. The total number of vehicle-km is known – 75 060 mill. vehicle km for passenger cars and 12.528 mill. vehicle km for trucks. Table 30: Numbers of cars and mill. vehicle km, the year 2002

Passenger cars 4,000,000 Trucks 330,000 Daytime Mio. vehicle km - cars 41,283 Daytime Mio. vehicle km - trucks 6,890 Source: KfV • Average 2002 price of fuel excluding tax and VAT. Table 31: Price of fuel excluding tax and VAT, the year 2002

Diesel € 0.316 Petrol € 0.293 Source: KfV

The calculation of extra fuel consumption due to DRL

A correction factor of 45% has been made for Austria. The correction is needed because 35% of car users already use DRL on a voluntary basis and it is assumed that 90% of car users will use DRL after the obligation. The value for the correction is then 0.55. Passenger cars – 41.283 mill. vehicle km / 50 km * 0.55 * 0.1 l = 45,000,000 litres Trucks – 6.890 mill. vehicle km / 50 km * 0.55 * 0.12 l = 9,100,000 litres

Page 69 DAYTIME RUNNING LIGHTS IN AUSTRIA

The fuel costs for cars and trucks are € 16,100,000 in 2002. In 12 years, the total cost is € 145,000,000.

5.3 Environmental effects

The cost of pollution arising as a result of DRL extra fuel emission is about 1% of the total costs of pollution caused by fuel emission in road transport [ETSC, 2003]. The estimated costs of road emissions in 2002 are € 2,232,385,000 [KfV]. Due to DRL use, the cost of € 200,000,000 is calculated for a 12-year period.

5.4 Calculation of other costs

5.4.1 Automatic light switch

The price of an automatic light switch in new cars is estimated at € 5. The number of new cars was 280,000 in 2002. The price of retrofitting amounts to € 50, including installation costs per old vehicle [ETSC, 2003]. The total cost for 12 years is € 44,000,000.

5.4.2 Maintenance and repair costs of automatic light switches during its lifetime

The costs are estimated to be € 15 per light switch [ETSC, 2003]. The total cost for 12 years is € 47,000,000.

5.4.3 Additional costs as a result of the wear of the bulbs during daytime use

• The replacement rate for bulbs increases by a factor of 2 due to DRL. The additional bulb costs are € 6 per car per year [ETSC, 2003]. The correction of 0.55 is needed. The total cost for 12 years is € 104,300,000.

Page 70 DAYTIME RUNNING LIGHTS IN AUSTRIA

6 Assessment Results

Table 32: Costs and benefits – 12-year period

Fatalities saved 1,040,000,000 € Extra fuel -145,000,000 € Emission costs -200,000,000 € Total benefits 695,000,000 € Light switch 44,000,000 € Maintenance 47,000,000 € Bulbs 104,300,000 € Total costs 195,300,000 € Cost / benefit 1/3.6

7 Decision-making process and barriers

In the mid 1990’s, the Austrian Road Safety Board [KfV] started its first awareness campaign for Daytime Running Lights (DRL) with information boards along streets with unusually high numbers of casualty accidents caused by passing. At that time, international studies in countries already using DRL indicated that about 30 fatalities could be saved in Austria every year due to such a safety measure. In 1996, the Federal Ministry of Transport launched a bill for a 2-year field test of DRL. During the following legislation process, several stakeholders voiced severe objections concerning additional fuel costs and stranded vehicles due to empty batteries. It was not till 2001 when the Austrian Road Safety Programme 2002-2010 was passed that DRL once again became a public agenda. In a comprehensive EU study based on the growing number of international studies, it was proven that DRL has a positive effect on reducing accidents. The introduction of daytime running lights in rural areas during winter was seen as a suitable way to overcome still existing concerns and objections. Besides the established safety effect of DRL, another aspect proved to be even more convincing. Most European countries have already established DRL, thus arguments finding a harmonized solution for the whole of Europe became more and more convincing. During a press conference in October 2004, the Austrian Minister of Transport, Hubert Gorbach, announced a new bill for DRL in the early months of 2005. Up to now, the main barrier in Austria preventing the obligatory use of daytime running lights consisted of objections from stakeholders (e.g. drivers’ unions) in technical and social aspects of the measure. Additional fuel consumption and the fear of elderly drivers getting stranded because of dead batteries have been major arguments in past discussions. Recent developments in vehicle technology (automatic switches for DRL) could dispel most of those objections.

REFERENCES

ETSC (2003): Cost Effective Transport Safety Measures

Page 71 CASE E1: four-arm roundabouts in urban areas In the czech republic

ROSEBUD WP4 – CASE E REPORT

FOUR-ARM ROUNDABOUTS IN URBAN AREAS IN THE CZECH REPUBLIC

BY PETR POKORNÝ

TRANSPORT RESEARCH CENTRE, CDV, THE CZECH REPUBLIC FOUR-ARM ROUNDABOUTS IN URBAN AREAS

TABLE OF CONTENTS

1 PROBLEM TO SOLVE ...... 75 2 DESCRIPTION...... 76 3 TARGET GROUP ...... 77 4 ASSESSMENT METHOD...... 77 5 ASSESSMENT QUANTIFICATION...... 79 6 ASSESSMENT RESULTS...... 80 7 DECISION MAKING PROCESS...... 80 8 CONCLUSION ...... 81

Page 73 FOUR-ARM ROUNDABOUTS IN URBAN AREAS

CASE OVERVIEW

Measure Implementation of four-arm roundabouts instead of four-arm intersections (without traffic lights) in urban areas (in cities with less than 100,000 inhabitants) Problem High number of accidents, high speeds Target Group All accidents at the treated sites Targets To reduce the number of accidents; traffic calming Initiator The initiator is mostly relevant local authorities, the owner of the infrastructure Decision-makers Members of city council, local authorities Costs Roundabout design costs and costs of implementation Benefits The only expected benefit is the reduction of accidents. Other impacts (on mobility and environment) were not calculated because of the lack of the available data.

Cost-Benefit Ratio 1/1.5

Page 74 FOUR-ARM ROUNDABOUTS IN URBAN AREAS

1 Problem

In the Czech Republic, more than 70% of accidents take place in urban areas and about 10% of them occur on four-arm intersections [Summary of Czech Accident Data, 2003]. One of the measures aimed at reducing the number of these accidents is to rebuild “dangerous” intersections into roundabouts. There are several reasons for implementation of roundabouts: their effects on improving road safety, on capacity, and on traffic calming. In some cases the roundabout can also be a significant architectural element of city design. The positive effects of properly designed and built roundabouts are well known from studies in many countries. In Czech traffic engineering, roundabouts are still quite a new element. In some cases there are still doubts on the use of roundabouts. Nevertheless, the number of roundabouts in the Czech infrastructure network is increasing (the quality of the design is problematic in some cases), but there are still a lot of barriers during the decision-making phase. There is not enough available data and studies evaluating the roundabouts in Czech infrastructure. One available source of information is the BESIDIDO project. It is a research project funded by the Ministry of Transport and elaborated by CDV and the Czech Technical University in ; its aim is to evaluate the affectivity of various infrastructure measures.

Figure 9: Number of road accidents on four-arm intersections in urban areas, 1999–2003

Number of accidents on four-arms intersection in urban areas (1999 - 2003)

17600 17409 17400

17200

17000

16947 16800

16726 16726 16695

number of accidents 16600

16400

16200

Source: CDV; Summary of Czech Accident Data 2003

Page 75 FOUR-ARM ROUNDABOUTS IN URBAN AREAS

Figure 10: Numbers of road accidents casualties on four-arm intersections in urban areas, 1999–2003

Number of casualties (1999 - 2003)

3500

2898 3000 2698 2938 2813 2840 2500

2000 fatalities seriously injured 1500 slightly injured

number of casualities of number 1000

360 347 360 378 500 338

49 50 4347 51 0 1999 2000 2001 2002 2003

Source: CDV; Summary of Czech Accident Data 2003

2 Description

2.1 General

Description of the sample There are eight roundabouts in the evaluated sample. All of them are four-arm roundabouts that were constructed instead of four-arm intersections between the years 1998–2002 in the urban areas of cities with population less than 100,000 inhabitants. Picture 1: Examples of roundabouts in the sample: Lázně Bohdaneč (left), Ždírec (right)

Source: CDV (project Besidido, 2004)

The brief description of the sample is in Table 33.

Page 76 FOUR-ARM ROUNDABOUTS IN URBAN AREAS

Table 33: Description of the sample

Site Number/City Population “ Before” Year of “After” Price (€) accident data implementation accident data 1.Česká Lípa 40,000 1995-1997 1998 1999-2000 unknown 2.Chlumec nad Cidlinou 5,000 2000 2002 2003 unknown 3.Chrudim 25,000 2000-2001 2002 2003 unknown 4.Lázně Bohdaneč 3,500 2000-2002 2003 2004 350,000 5.Litomyšl 10,000 1999 - 2000 2001 2002 unknown 6.Most 70,000 1999 2000 2001-2003 200,000 7.Tábor 37,000 1996-1997 1998 1999-2000 unknown 8.Ždírec 3,000 2000-2001 2002 2003-2004 unknown Source: CDV (Project Besidido, 2004) All roundabouts in the sample are “typical“ four–arm roundabouts designed in accordance with Czech technical standards. The reason for their implementation was mostly the demand for more capacity and for improving the safety situation.

3 Target Group

The implementation of a roundabout has mostly a positive effect on the safety level of the treated site. This is based on the fact that the roundabout geometry reduces the number of collision points, decreases the speed of vehicles, and improves the safety of pedestrian crossing. The only negative phenomenon is a possible lower safety level for cyclists. Therefore, the target accident group was defined as “all accidents occurring on the treated sites”. The sample contains 8 sites, where the original four-arm intersections without traffic lights were rebuilt into the four-arm roundabouts. Based on accident data before the implementation of roundabouts, an “average” intersection accident is determined. This is an accident with 0.004 fatality, 0.04 severely injured, 0.19 slightly injured and with property damage valued at 27,000 CZK. The value of one average accident is calculated to be € 7,500 (at 2002 prices) [based on the socio-economic evaluation of road accident; Koňárek, 2003].

4 Assessment method

The ideal method of assessment would be to provide the complete CBA (with calculation of roundabout effects on environment and mobility). The quality of available data does not allow for such a complete analysis, so only the safety effects are calculated in the analyses. The suitable method for such calculation is a method combining before/after comparison with a control group of sites (sites which are similar in most characteristics to the treatment sites, but left untreated). In this calculation, the total number of accidents on four-arm urban intersections in the whole country is used as a control group, so the general trends in accident number development are taking into account. The aim of the calculation is to find the number of accidents prevented by the implementation of roundabouts instead of four-arm intersections in the evaluated sample

Page 77 FOUR-ARM ROUNDABOUTS IN URBAN AREAS of eight sites. The “before” and “after” accident data of treated sites and of all four-arm intersections in the Czech Republic were known.

An evaluation of the treatment effect θi at each site by means of the odds-ratio with the comparison group is calculated. A correction due to changes in traffic volumes is not performed, so δ = 1. The formula is: X Estimated effect(θ ) = a δ Ca X m Cb whereh

Xa – the number of accidents observed at the treatment site in the “after” period,

Xm – the number of accidents at the treatment site in the “before” period,

Ca – the number of accidents in comparison group sites in the “after” period,

Cb – the number of accidents in comparison group sites in the “before” period,

Weighting the effects found for separate treatment sites is done by means of a standard method for weighting odds-ratios, where a statistical weight of separate result is defined by the sizes of data sets, which provided the following result:

∑ wi ln(θi ) Weighted mean effect(WME) = exp( i ) ∑ wi i 1 1 wi = = VAR(log(θ i )) 1 1 1 1 i + i + i + i X a X b C a C b where

θi - estimate of effect for site i, wi - statistical weight of estimate for site i, i X a – the number of accidents observed at treatment site i, in the “after” period, i X b – the number of accidents at treatment site i, in the “before” period, i C a – the number of accidents in comparison group (for site i), in the “after” period, i C b – the number of accidents in comparison group (for site i), in the “before” period. The 95% confidence interval for the weighed effect is estimated as follows:      z α z α     1−  WME exp 2 ,WME exp 2    w   w    ∑ i   ∑ i    i   i  The applicable value of the safety effect, i.e. the best estimate of accident reduction associated with the treatment (in percent), is calculated as (1-WME)*100.

Page 78 FOUR-ARM ROUNDABOUTS IN URBAN AREAS

5 Assessment Quantification

The unit of implementation A four-arm roundabout was determined to be the typical unit of implementation. The typical cost of the unit of implementation The typical cost was estimated to be € 300,000 (at 2002 prices). The estimate was based on results found in the BESIDIDO project. The cost of maintenance was not calculated due to an assumption that the cost of maintenance is similar for four-arm intersections as it is for the four-arm roundabout. The duration of the effect The duration of the effect was estimated to be 20 years. The discount rate The discount rate was determined to be 5%. This is based on the recommended value of discount rate used in the Rosebud project. All prices are converted to Euro; the price level is as of the year 2002. Price of a typical four-arm intersection accident The price of a typical four-arm intersection accident was calculated to be € 7,500 (at 2002 prices). The calculation is based on accident statistics of the intersections from the sample before the implementation of roundabouts.

5.1 Safety effect

The aim was to find the number of accidents, which will be prevented by the implementation of roundabouts instead of four-arm intersections, in an evaluated sample of eight sites. Table 34: Data for calculations site site accidents comparison group estimated statistical weight number before after before after effect θi of estimate wi 1 85 24 57810 34356 0,475 18,699 2 5 5 17409 16695 1,04 2,5 3 36 3 34135 16695 0,17 2,768 4 13 5 50861 16600 1,178 3,61 5 2 1 34356 16726 1,027 0,666 6 10 4 16947 50147 0,135 1,428 7 27 29 38810 34356 1,213 13,971 8 19 1 34135 33295 0,054 0,949

Table 35: Safety effect of evaluated roundabouts Estimated effect (WME) WME confidence Number of treatment Number of accidents at interval sites in the sample the treatment sites 0.624 (0.465, 0.836) 8 197

The average accident reduction associated with the treatment is calculated as (1-WME) x 100 = (1- 0,0,624) x 100 = 37.6%.

Page 79 FOUR-ARM ROUNDABOUTS IN URBAN AREAS

Table 36: Accident reduction

Site Average annual Reduction of number number of accidents accidents

1 28.3 10.64 2 5 1.88 3 18 6.77 4 4.3 1.62 5 1 0.37 6 10 3.76 7 13.5 5.08 8 9.5 3.57

The total sum of accidents prevented annually multiplied by the average accident costs (the total benefit) is 33.7 x 7,500 = € 253,000. The annual average sum of money saved for one treated site is € 31,625.

6 Assessment Results

The total cost of prevented accidents in a period of 20 years at one treated site is calculated to be € 444,000. Because the cost of one unit of implementation is estimated at € 300,000, the cost/benefit ratio is 1/1.5. Table 37: Costs and benefits – 20-year period

Accidents prevented € 444,000 Cost of one unit € 300,000

Cost / benefit 1/1.5

7 Decision-Making Process

The cost-benefit calculation of the roundabout implementation in urban areas is not a common tool in decision-making processes in the Czech Republic (it has probably never been used). The decisions regarding implementation of roundabouts are usually made by the relevant local authority, which is the owner of the urban infrastructure. The criteria for decisions and implementations are mostly as follows: • Traffic engineering – capacity issues, traffic calming • Safety of all road users • Town planning It is generally agreed among the experts and decision-makers that roundabouts are a “safe” type of intersection. The fundamental arguments against their implementation are mostly based on the general feeling of decision-makers that the capacity of roundabouts is rather limited. The reason for it could be the fact that some of the already-implemented roundabouts have been causing traffic congestions, with obvious impacts on mobility and environment. Wrong roundabout design mostly causes these problems.

Page 80 FOUR-ARM ROUNDABOUTS IN URBAN AREAS

The CBA, which would compare the safety effects of roundabouts with their effects on environment and mobility, could thus be a very useful tool to improve the decision-making process.

8 Conclusion

Due to the limited sources of available data, it was not possible to calculate a complete CBA. A “mini-CBA” was thus calculated - only the safety effects of roundabouts were taken into account. The effects on environment and mobility were not taken into account. The result showed that the four-arm roundabouts in urban areas have a positive effect (-37.6%) on the reduction of all accidents.

REFERENCES

The Czech Statistical Office: www.czso.cz The Czech Ministry of Interior: http://www.mvcr.cz/statistiky/crv.html Koňárek (2002): Socio-economy losses caused by accidents in CZ The Traffic Police Directorate of CZ (2003): Summary of accidents data WP3 (2004): Improvements in efficiency assessment tools, ROSEBUD

Page 81 CASE E2: Speed humps on local streets

Technion - Israel Institute of Technology Transportation Research Institute

ROSEBUD WP4 - CASE E REPORT

SPEED HUMPS ON LOCAL STREETS

BY VICTORIA GITELMAN AND SHALOM HAKKERT,

TRANSPORTATION RESEARCH INSTITUTE, TECHNION, ISRAEL SPEED HUMPS ON LOCAL STREETS

TABLE OF CONTENTS

1 THE PROBLEM TO SOLVE...... 85 2 DESCRIPTION OF MEASURE...... 85 2.1 General ...... 85 2.2 Current installation ...... 87 3 TARGET ACCIDENT GROUP...... 88 4 ASSESSMENT TOOLS ...... 88 4.1 Method for estimating safety effect ...... 88 4.2 Safety effect of speed humps...... 90 4.3 Accident costs...... 91 5 COST-BENEFIT ANALYSIS...... 92 5.1 General ...... 92 5.2 Values of costs and benefits ...... 92 5.3 Cost-Benefit Ratio ...... 93 6 DECISION MAKING PROCESS...... 93 7 DISCUSSION...... 94

Page 83 SPEED HUMPS ON LOCAL STREETS

CASE OVERVIEW

Measure Installation of speed humps on a section of urban street Problem High travel speeds along the road section and accident occurrences Target Group All injury accidents on the treated road Targets Reducing travel speeds and the number of injury accidents along the road Initiator Local authorities – for the measure’s application; Ministry of Transport – for the evaluation of safety effect Decision-makers Local authorities Costs Speed humps’ design and installation costs, paid by the local authority Benefits The benefits are expected from the savings in injury accidents along the treated road. The costs of time losses due to lower vehicle speeds are subtracted from the benefits. The residents of the area and the national economy will benefit from the measure’s application. Cost-Benefit Ratio May range from 1:4 to 1:2, depending on the type of speed humps installed.

Page 84 SPEED HUMPS ON LOCAL STREETS

1 Problem

In Israel, similar to many other countries, more than 70% of injury accidents and about half of fatal accidents occur in built-up areas (Gitelman, Hakkert, 2003). Previous research indicates that, as to the location of road accidents in towns, there is a somewhat equal subdivision of those occurring on arterials, and in central city districts and residential areas. Following this, the number of injury accidents in the residential areas throughout the country amounts to some 5,000 per year, with 9,000 injuries involved. Due to the scattered pattern of accidents in residential areas, on the one hand, and the high proportion of vulnerable road users on the residential streets, on the other hand, traffic calming is known as the best safety solution for such areas. Safety effects of traffic calming measures stem mostly from reduced travelling speeds and also from a reduction in traffic volumes on residential streets. Traffic calming measures are engineering solutions that change the regular road layout. These measures can be subdivided into two groups: those creating a horizontal diversion from a regular road lane and those creating a vertical diversion from a regular road surface. The latter group includes speed humps. Speed humps may serve as one of the design elements when a traffic calming area ("30- km zone") is established. In this case, speed humps are usually combined with other measures, e.g. road narrowings, chicanes, pedestrian refuges and roundabouts. Regarding the maintenance and improvement of existing roads, speed humps are frequently applied by the authorities when the street design does not satisfy safety demands, i.e. when actual vehicle speeds are higher than they should be for the given road type and surroundings, or when road accidents occurred on the street or in the area considered. Sometimes a demand for the installation of speed humps comes from the residents, who are worried about the high travel speeds or of near-accidents that were observed on the street. Speed humps are frequently chosen as a typical solution when there is a need to reduce travel speeds on a local street and to provide the street with a calmer and safer character.

2 Description of measure

2.1 General

Speed humps are defined as raised areas over the road surface, which are installed over the whole road width or part of it, and present a physical measure for reducing travel speeds (Guidelines, 2002). The humps consist of a raised road pavement and can be made of asphalt, concrete or paving blocks. The main advantages of speed humps are in their self-enforcing nature and in creating a visual impression that the street is not designated for high speeds or for passing traffic (e.g. ITE, 1997). Over the last three decades, safety effects of speed humps were examined and proven in many countries. Those are associated with two basic reasons: typically, a reduction in travel speeds and, frequently, a reduction in traffic volume, following the humps' installation. The safety effect is usually observed provided that the installation parameters

Page 85 SPEED HUMPS ON LOCAL STREETS and the density of the humps are proper, i.e. strict enough in order to dictate the desired travel speeds on the street. The speed humps' installation may have one of two purposes (Guidelines, 2002): a) reducing travel speeds along a road section; b) reducing travel speeds near a problematic point, e.g. a pedestrian crossing, a school, or another public place with a high concentration of pedestrians. The first case is considered as the typical one and demonstrating major advantages of the measure. Speed humps are known in the world since 1973, when the first systematic study aimed at developing speed humps was conducted in the UK (Watts, 1973). The first humps had a circular profile and, until today, it is the most widespread form of speed hump in many countries. Several years later, another form of speed humps - a trapezoidal profile, was independently developed in two countries: Australia and the USA. While a circular hump resembles a segment of a circle, a trapezoidal hump consists of three components: an incline ramp, a flat head and a decline ramp. Figure 11 illustrates typical parameters of circular and trapezoidal humps, which are called using their historical names: "Watts profile" for a circular hump (after the name of the researcher who developed the first humps), and "Seminole profile" for a trapezoidal hump (after the name of the county in Florida, USA, where the humps were developed). The circular and trapezoidal humps are the basic (regular) types of speed humps that are in use today around the world. Over the last decades, many variations of basic humps were developed in the UK, the Netherlands, Denmark, Germany and other countries (e.g. Gitelman et al, 2001). Among other types, speed cushions (narrow trapezoidal humps allowing for easy passing by buses and large vehicles), sinusoidal profile humps and combi-humps (a combination of speed cushion and regular humps) were introduced and tested in some European countries. In Israel, the updated edition of guidelines for design and installation of speed humps was published by the Ministry of Transport in 2002 (Guidelines, 2002). The types of speed humps that are recommended for the use in urban areas in Israel are: 1. Circular humps, of 3.5-4 m in length, with a height of 8-10 cm for a street with a 30 kph speed limit and a height of 6-8 cm for a 50 kph speed limit; 2. Trapezoidal humps, with a height of 8-10 cm for a 30 kph speed limit and a height of 6-8 cm for a 50 kph speed limit. The flat head of the humps should be of 2.5-3 m in length and the slope of the ramps not steeper than 1:10- 1:15. 1. Speed cushions, for the streets with a 50 kph speed limit. These should be 6- 8 cm in height, 1.9-3.7 m in length, and 1.6-2.0 m in width. The slope of the incline/ decline ramps should be 1:8-1:10.

Page 86 SPEED HUMPS ON LOCAL STREETS

Figure 11: Basic profiles of speed humps in the historical perspective: circular (Watts) and trapezoidal (Seminole).

Sources: Ewing (1999); Weber and Braaksma (2000).

2.2 Current installation

In the current study, we consider the installation of regular speed humps (i.e. circular or trapezoidal humps) on a typical urban street with a 50 kph speed limit. The road section for the treatment is about 500 m in length. To note, according to the Guidelines (2002), 500 m is the maximum recommended length of road section which can be treated by continuous speed humps only, whereas a longer road section needs a combination of speed humps with other traffic calming measures. For the street with the 50 kph speed limit the parameters of speed humps can be as follows: Circular hump – 8 cm in height, 3.7 m in length; Trapezoidal hump – 8 cm in height, the flat head of 2.5 m in length, slopes of 1:10, total length of 4.1 m. The purpose of the installation of speed humps on the road section is to provide that the level of actual speeds (85%) be below the speed limits (50 kph). Based on the known relationships between the density of speed humps and the actual travel speeds along the road (Guidelines, 2002), the recommended distances between the humps considered should be 100-130 m for circular humps and 90-110 m for trapezoidal humps. Therefore, over the road section considered, five speed humps should be installed.

Page 87 SPEED HUMPS ON LOCAL STREETS

3 Target Accident Group

Considering the speed humps' installation, the safety effect usually refers to all injury accidents (e.g. Webster and Layfield, 1996). This is based on the assumption that reducing actual speeds creates a moderating effect on all accident types, i.e. single- vehicle accidents, multiple-vehicle collisions and pedestrian accidents. Therefore, estimating a safety effect of speed humps' installations on urban roads in Israel, the target accident group was defined as all injury accidents on the treated roads. A slightly different consideration is accepted when a single site is considered for a speed hump installation. For example, according to Guidelines (2002), a warrant for the installation of speed humps suggests to account for a weighted number of accidents, where a severe accident of any type has the weight of 5; a pedestrian accident – the weight of 1; other accidents – weights of 0.5. Such an approach was chosen in order to strengthen the consideration of the speed factor in accident occurrences. Examining the warrant, the accident numbers for the last 3-5 years are weighted and an average annual number is considered. On the urban street considered in this study, three injury accidents occurred over the three last years, of which one was a pedestrian accident and two were vehicle collisions; all accidents produced slight injuries. Using the warrant's approach, the weighted number of accidents on the street of treatment will be 1 * 1 + 2 * 0.5 = 2 injury accidents in 3 years, or 0.67 accidents per year.

4 Assessment tools

4.1 Method for estimating safety effect

The safety effect from the installation of speed humps on urban roads in Israel was estimated in a recent study, which was initiated by the Ministry of Transport and conducted by the T&M Company in association with the Technion (Hakkert et al, 2002). The study aimed at developing a uniform methodology for evaluating potential safety effects of projects on road infrastructure improvements and estimating safety effects of some 30 types of safety treatments, which were introduced on Israeli roads through the 90s. For the estimation of safety effects of road infrastructure improvements, a method combining an after/before comparison with a control group with an empirical correction due to selection bias, was proposed. The outline of the method resembles that described in Elvik (1997), whereas in the Israeli study an extension accounting for changes in traffic volumes was developed. Besides, the reference group statistics that are necessary for correction of the selection bias were estimated by the method of sample moments and not on the basis of a regression model. The reference group included sites which are similar to the treatment sites in most engineering characteristics but left untreated (unchanged) during the “before” periods of all the sites in the treatment group. The demands for the control (comparison) group were as follows: it should be large (to strengthen the significance of the findings) and demonstrate some similarity with the treatment group from the engineering viewpoint. For a treatment type considered, evaluation of the safety effect included three steps: 1) A correction of “before” accident numbers with the help of reference group statistics for each site in the treatment group (WP3, 2004 – see Appendix to Chapter 3).

Page 88 SPEED HUMPS ON LOCAL STREETS

2) An evaluation of the treatment effect at each site by means of the odds-ratio with the comparison group, where for the “before” period the corrected accident numbers (from the first step) are applied. Besides, a correction due to changes in the traffic volumes is performed. The formula has the form: X Estimated effect(θ ) = a δ Ca X m Cb where 1 δ = βc βt Vcb  Vta      Vca  Vtb  where

Xa – the number of accidents observed at the treatment site in the “after” period,

Xm – the corrected number of accidents at the treatment site in the “before” period,

Vta – traffic volume at the treatment site in the “after” period,

Vtb – traffic volume at the treatment site in the “before” period,

Ca – the number of accidents in comparison group sites in the “after” period,

Cb – the number of accidents in comparison group sites in the “before” period,

Vca - traffic volume in comparison group sites in the “after” period,

Vcb - traffic volume in comparison group sites in the “before” period,

βt – the parameter of safety performance function (a power of relation between traffic volume and the accident number), for treatment sites,

βc – the parameter of safety performance function, for comparison-group sites. 3) Weighting the effects found for separate treatment sites. This is done by means of a standard way known for weighting odds-ratios, where a statistical weight of separate result is defined by the sizes of data sets, which provided this result:

∑ wi ln(θi ) Weighted mean effect(WME) = exp( i ) ∑ wi i 1 1 wi = = VAR(log(θ i )) 1 1 1 1 i + i + i + i X a X b C a C b where

θi - estimate of effect for site i, wi - statistical weight of estimate for site i, i X a – the number of accidents observed at treatment site i, in the “after” period, i X b – the number of accidents at treatment site i, in the “before” period, i C a – the number of accidents in comparison group (for site i), in the “after” period, i C b – the number of accidents in comparison group (for site i), in the “before” period. Page 89 SPEED HUMPS ON LOCAL STREETS

The 95% confidence interval for the weighed effect is estimated as follows:      z α z α     1−  WME exp 2 ,WME exp 2    w   w    ∑ i   ∑ i    i   i 

The applicable value of the safety effect, i.e. the best estimate of accident reduction associated with the treatment (in percent), is calculated as (1-WME)*100. In the cases of large samples of treatment sites (that diminishes a threat of selection bias and also limits the practical possibility of building a comparable reference group), only steps 2-3 were applied for the evaluation.

4.2 Safety effect of speed humps

In the study Hakkert et al. (2002), the data on the road infrastructure improvements were collected by means of written applications and meetings with the representatives of road and municipal authorities in different country areas. A special database on the issue was established. The data were sought mostly on projects performed in the mid 90s, to have a two-year “before” and two-year “after” period for observation. To represent a specific project in the database, three information elements were defined as crucial: site of treatment, type of treatment and the period of treatment. For the project to be involved in the evaluation, all three pieces of information had to be thoroughly verified. To provide a minimum but comprehensive presentation of a specific project in the database, a special reporting form was devised which enabled to classify the site and the treatment in accordance with the road layout, area specifics, etc. The data were obtained from the authorities and accomplished by information from detailed maps, field surveys and the publications of the Central Bureau of Statistics (CBS). Within each treatment type for the analysis, a strict definition of the periods “before” and “after” the treatment was provided for each site; a relevant definition of both periods for the comparison-group sites was also attached. The next stage in data preparation was filtering the CBS accident files for the sites and periods required. For each treatment type, files with series of accident numbers were produced for every treatment and comparison group of sites and then processed using the method described in Section 4.1. For the treatment type "installation of speed humps on a local street", the data were collected on the majority of projects, which were performed by 3 municipalities: Tel-Aviv, Netanya and Haifa. Over the years 1994-1998, speed humps were installed on 94 streets of these towns. The time period for the consideration was 1991-1999, both for the treatment and comparison group roads. For the treatment roads, all injury accidents observed on these roads were considered, whereas for each treated street a two-year "before" period and a two-year “after” period were separately defined. All injury accidents observed on urban road sections throughout the country (excluding junctions and fitting "before" and "after" periods for each site of treatment) served as a comparison group. Table 38 details the number of sites (projects) involved in the evaluation, the number of accidents observed at the treatment sites in “before” and “after” periods, the mean value of the safety effect estimated and the confidence interval for this value. As can be seen from Table 1, a significant accident reduction was observed following the treatment. (A reduction is significant when the whole WME confidence interval is below one.) Page 90 SPEED HUMPS ON LOCAL STREETS

Table 38: Safety effect of speed humps estimated for Israeli conditions Treatment type Estimated WME Number of Number of effect confidence treatment sites accidents at the (WME) interval in the sample treatment sites Speed humps on 0.603 (0.44, 0.828 ) 94 129 urban road sections Source: Hakkert et al, 2002 The average safety effect of speed humps installed on urban roads in Israel was a 40% reduction in injury accidents. This result is comparable with the international value reported by Elvik et al (1997) – a 48% reduction in injury accidents.

4.3 Accident costs

In the current Israeli practice, the average accident cost is estimated as a sum of injury costs and damage costs of an average accident in the target accident group. The injury costs are a sum of injury values multiplied by the average number of injuries, with different severity levels, which were observed in the target accident group. The road accident injury values are usually taken as $ 500,000 per fatality, $ 50,000 per serious injury, $ 5,000 per minor injury; the damage value is stated as 15% of the injury costs (Guidelines, 2002). Table 39 illustrates the calculation of accident costs for an average injury accident observed on urban Israeli roads over the period 1996-2000. The injury costs of an average accident are NIS 77,490; with the addition of damage-costs, the value of average injury accident is NIS 89,114 (at 2000 prices). The above values of injury should be treated as conservative because a recent evaluation of losses from road accidents in Israel recommended a higher estimate of the fatality value of $ 930,000 (MATAT, 2004). The latter accounts for both lost output and human costs, i.e. applies the willingness-to-pay approach. Table 39: Estimating costs for an average injury accident on urban Israeli roads

Value Fatality Serious injury Minor injury Average number of injuries per 0.01 0.11 1.59 accident* Injury-values, $ 500,000 50,000 5,000 Total injury-costs of average $ 18,450 or NIS 77,490 accident** Damage costs NIS 11,624 Total costs of an average accident NIS 89,114 (at 2000 prices) *over the period 1996-2000 **$ 1 = 4.2 NIS

Page 91 SPEED HUMPS ON LOCAL STREETS

5 Cost-Benefit Analysis

5.1 General

In this section, a Cost-Benefit Analysis (CBA) of the installation of speed humps on a local street is performed. The CBA compares the measure's benefits with the measure's costs, where both values are brought to the same economic framework. The main benefit from the installation of speed humps stems from the accident reduction that is expected after the treatment. However, due to a reduction in vehicle speeds that will be attained on the treated road, a loss in travel time by the vehicles passing the road should be accounted for, too. The economic value of the time lost should be subtracted from the value of benefits. The costs of the measure are a direct result of the initial investment, which is required for the design and installation of speed humps along the street considered. No special maintenance expenses are required as this is supposed to be a part of regular road maintenance. Both the costs and benefits are considered for 5 years, with a 7% discount rate; the accumulated discount factor (D) is 4.10.

5.2 Values of costs and benefits

The cost of speed humps' installation should account for the expenses on: the hump's design and its approval process, a dismantling of the road surface, building the hump, road signing and marking. When more than one unit of speed humps is installed, the unit cost times the number of the installed units should be taken into account. Using the typical cost values of the regular speed humps, which are provided by the Guidelines (2002) and the Israeli study of the road infrastructure improvements – Hakkert et al (2002), the cost value of one unit may range from 3,000 to 6,000 NIS (NIS – New Israeli Shekel). Therefore, the costs of installation of speed humps on the street considered will be NIS 15,000-30,000 (at 2000 prices). The one-year value of benefits from the expected accident reduction is estimated as a product of the annual number of "before" accidents, the accident reduction factor (the safety effect) and the accident cost. This value is: 0.67 accidents * 0.4 * 89114 NIS/ accident = 23,883 NIS (at 2000 prices). The one-year value of time losses due to the humps' installation is estimated as a product of the time lost by one vehicle, the average daily traffic volume, the time costs and the number of working days over the year. Comparing the time required for a vehicle to pass the street with a higher speed (before the humps' installation) with the time required to pass the same street with a lower speed (after the humps' installation), one can conclude that the average delay will be of 4 sec/vehicle. (To note, a similar value was provided by Atkins and Coleman (1997), who measured the values of time lost by one vehicle due to a regular hump and found that even for large vehicles it is 1 sec per hump, on average.) The daily traffic volume on the street of treatment is 8000 vehicles. The cost of a delay of an average vehicle on a local street can be estimated as 3.96 NIS/hour (as some 20% of typical costs of delay for the economy - see Guidelines, 2002). Over the year, there are

Page 92 SPEED HUMPS ON LOCAL STREETS

260 working days (52 weeks * 5 working days); only working days are considered for time costs, so weekends may be neglected. Therefore, the one-year value of time lost due to the humps' installation on the street considered is: 4 sec/vehicle * 1/3600 hours * 8000 vehicles * 3.96 NIS/hour * 260 days = 9,152 NIS (at 2000 prices).

5.3 Cost-Benefit Ratio

Table 40 illustrates the calculation of the cost-benefit ratio (CBR) of the speed humps' installation. The value of the measure's costs is 15,000-30,000 NIS (at 2000 prices) or 3,600-7,200 Euro (at 2002 prices). The total value of benefits is calculated as the difference between the costs of accidents prevented and the costs of time losses, multiplied by the accumulated discount factor (D = 4.10). The total value of benefits is 60,397 NIS (at 2000 prices) or 14,408 Euro (at 2002 prices). Depending on the measure's costs, the CBR ranges from 1:4 to 1:2. For the local street considered, the installation of speed humps appears to be cost- effective. Table 40: Calculation of the cost-benefit ratio

Costs Benefits Costs of Losses: Costs accidents of vehicle prevented in delays in one one year, year, NIS NIS Costs of one speed hump, NIS 3,000-6,000 23,883 -9,152 Total benefits in one 14,731 year, NIS Costs of a series of 5 humps, 15,000- Total benefits in 5 60,397 NIS (2000) 30,000 years, NIS (2000) Total costs, Euro (2002)* 3,578-7,156 Total benefits, Euro 14,408 (2002)* Cost-benefit ratio From 1:4.0 to 1:2.0 *Change of price index over 2000-2002 is 1.0687. In 2002: 1 Euro = 4.48 NIS.

6 Decision-Making Process

The cost-benefit analysis of the installation of speed humps is not common in Israel as this treatment is considered by local authorities as a low-cost measure and therefore, generally does not require an economic justification. As a result of more than 20 years of practical experience with their application, the safety effect of speed humps is widely accepted by the professional community and the local authorities. Typical questions usually concern the humps’ installation parameters and the suitability of the measure to the road’s layout, and much less – the economic effect of the measure.

Page 93 SPEED HUMPS ON LOCAL STREETS

Besides, pressure to install speed humps sometimes comes from the residents of the area who are interested in calming the traffic and in preventing accidents that might occur. Being under public pressure, the authorities feel they do not need an economic evaluation to promote the measure’s application. On the contrary, the economic evaluation of the speed humps’ installation might sometimes be helpful to demonstrate the lack of efficiency of the measure considered, allowing to rank the sites to be treated and the measures to be applied.

7 Discussion

In this study, a CBA of a typical example of a speed humps’ installation on an urban street was considered. The measure was found to be beneficial, mostly due to the fact that injury accidents were observed on the road in the “before” period. The economic consideration accounted for the humps’ installation costs, the safety effect expected and the costs of time losses due to lower travel speeds. The environmental impact of the measure, e.g. changes in the level of pollution or noise over the street, was not considered, as it is not essential in such a kind of installation. For instance, as indicated by different studies (Gitelman et al, 2001), the positive and negative pollution effects of speed humps usually compensate each other, especially where the parameters and the density of their installation are proper (i.e. keeping a certain speed level over the whole road section). The safety effect of speed humps was significant under Israeli conditions, in line with the findings reported by studies in other countries. The current study accounted for the time losses due to speed humps, which does not present a common component in the economic evaluation of this measure. One should remember that under certain conditions (e.g. for a road with higher traffic volume) the measure not be beneficial. The CBA presented in this study can be characterized as follows: • the evaluation findings support the measure's implementation; • to estimate the safety effects, statistical models were fitted to the accident data, and the evaluation was in line with the criteria of correct safety evaluation (WP3, 2004); • the accident costs were fitted to the accident type considered, however, they should be treated as conservative as the injury costs did not account for the willingness-to-pay component; c) the evaluation of the safety effect was initiated by the Ministry of Transport. However, the CBA of the measure was not required by the decision-makers.

Page 94 SPEED HUMPS ON LOCAL STREETS

References

Atkins C. and Coleman, M. (1997) The influence of traffic calming on emergency response times. ITE Journal, August, pp. 42-46. Gitelman V., Hakkert A.S. et al (2001). Speed humps in towns. A literature survey. Ami- Matom Company and the Technion, Haifa (in Hebrew). Gitelman V., Hakkert A.S. (2003). A wide-scale safety evaluation of traffic calming measures in residential areas. European Transport Conference, Strasbourg, France. Guidelines (2002). Design and performance of speed humps. Ami-Matom Company, Ministry of Transport (in Hebrew). Elvik, R. (1997). Effects on Accidents of Automatic Speed Enforcement in Norway. Transportation Research Record 1595, TRB, Washington, D. C., pp.14-19. Elvik, R., Borger-Mysen, A. and Vaa, T. (1997) Trafikksikkerhekshandbok (Traffic Safety Handbook). Institute of Transport Economics, Oslo, Norway. Ewing, R. (1999) Traffic Calming. State of the Practice. Federal Highway Administration, US Department of Transportation, and Institute of Transportation Engineers, Washington, DC. Hakkert, A.S., Gitelman, V., et al (2002) Development of Method, Guidelines and Tools for Evaluating Safety Effects of Road Infrastructure Improvements. Final report, T&M Company, Ministry of Transport (in Hebrew). ITE (1997). Guidelines for the Design and Application of Speed Humps. A recommended practice of the Institute of Transportation Engineers, Publication No. RP-023A, Washington, DC. MATAT (2004). Road Accidents in Israel: the scope, the characteristics and the estimate of losses to the National Economy. MATAT - Transportation Planning Center Ltd, Ministry of Transport. Weber, P.A. and Braaksma, J.P. (2000). Towards a North American Geometric Design Standard for Speed Humps. ITE Journal, January. Webster, D. and Layfield, R. (1996). Traffic calming – Road hump schemes using 75mm high humps. TRL Report 186, Transport Research Laboratory, Crowthorne, UK. WP3 (2004). Improvements in efficiency assessment tools. ROSEBUD.

Page 95 CASE E3: TRAFFIC CALMING MEASURES

National Technical University of Athens Department of Transportation Planning and Engineering

ROSEBUD WP4 - CASE E REPORT

TRAFFIC CALMING MEASURES

IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

BY GEORGE YANNIS AND PETROS EVGENIKOS

NTUA / DTPE, GREECE IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

TABLE OF CONTENTS

1 PROBLEM ...... 99 2 DESCRIPTION...... 99 2.1 Speed humps and woonerfs description ...... 99 2.2 Description of areas where traffic calming measures were implemented...... 101 3 TARGET GROUP ...... 102 4 ASSESSMENT METHOD...... 102 4.1 General ...... 102 4.2 Estimation of safety effect ...... 102 5 ASSESSMENT QUANTIFICATION...... 105 5.1 Traffic calming measures implementation cost ...... 105 5.2 Traffic calming measures benefits...... 105 5.2.1 Number of accidents prevented ...... 105 5.2.2 Accident cost...... 107 5.2.3 Estimation of cost for time lost ...... 108 6 ASSESSMENT RESULTS...... 109 7 DECISION MAKING PROCESS...... 110 8 IMPLEMENTATION BARRIERS ...... 110 9 CONCLUSION / DISCUSSION...... 111

Page 97 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

CASE OVERVIEW

Measure Implementation of low cost road traffic engineering measures (speed humps and woonerfs) in one direction - one-lane roads in the Municipality of Neo Psychiko in the Greater Athens Area in Greece Problem to solve In Greece 72% of the total number of road accidents occur in urban areas and speed is the most significant factor leading to their continuous increase. Increased travel speeds along urban roads affect not only the road accident causation, but also the accident severity. Target Group Inhabitants of residential areas (pedestrians, children, two-wheelers, drivers, passengers) Targets a) Creation of calm driving areas b) Decrease in the number of road accidents and related casualties Initiator Municipality of Neo Psychiko, Ministry of Public Works Decision-makers Municipality of Neo Psychiko. Costs Implementation costs (design and installation/construction) for speed humps and woonerfs provided by municipal funds from the Municipality of Neo Psychiko. Benefits: Fatal and injury accidents prevented Cost/Benefit Ratio 1:1.14 to 1:1.2

Page 98 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

1 Problem

In Greece, less than 1,600 persons killed and 19,000 persons injured are recorded in more than 16,000 road accidents annually (DTPE, 2004). More specifically, 76% of the total number of road accidents occurs in urban areas. Speed is the most significant factor leading to the high increase of road accidents. High speed is a major factor in road accidents, as it affects both their occurrence and their severity (KANELLAIDIS et al, 1995). The majority of Greek drivers exceed the speed limit in urban areas, and therefore road accidents in urban areas present a continuously increasing trend (KANELLAIDIS, et al, 1999). There are a wide variety of methods and techniques used for reducing road accidents in urban areas, such as enforcement, intensive campaigns, specific traffic management techniques etc. However, Low Cost Traffic Engineering Measures (LCTEM) (or traffic calming measures) are deemed to be the most efficient measures towards tackling one of the most significant problems that communities face nowadays: urban road accidents.

2 Description

2.1 Speed humps and woonerfs description

Speed humps are raised paved areas on the surface of road, extended across its width. They are constructed by different types of materials, as asphalt, concrete, bricks or plastic (caoutchouc) and are usually designed for travel speeds between 20 - 30 km/h [KAPICA C.J, 2001]. Their length is usually larger than the distance between the wheels of vehicle (usual length 3.6 m), their height oscillates between 7.5 - 10 cm and the recommended distance between successive humps varies from 60 to 100 m. (ZAIDEL et al, 1992). The main advantages and disadvantages deriving from the use of speed humps in the road network of an urban area are shown in the following Table 41. Table 41: Advantages and disadvantages of speed humps Advantages Disadvantages

1. Decrease of the number of conflicts of 1. Obstruct the movement of heavy vehicles at junctions vehicles 2. Travel speed reduction 2. Require additional traffic signing 3. Do not prohibit the movement of 3. Create potential deviation of traffic in vehicles near roads 4. Provide aesthetics of environment for 4. Influence traffic islands pedestrians and pedal cyclists 5. Require maintenance 5. Positive effects in multi-sectoral nodes 6. Low construction costs

Source: Jacksonville Florida City, 2000 The implementation of speed humps in several developed countries resulted in considerable improvement of road safety at the local level. In Denmark, a reduction in road accidents and road casualties by 24% and 45% respectively was attributed to the introduction of such traffic calming measures (ENGEL, THOMSEN, 1992). Some types of speed humps that can be used at urban areas are presented in Figure 12.

Page 99 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

Figure 12: Typical dimensions of basic types of speed humps

Source: ZAIDEL et al, 1992

Page 100 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

Woonerfs, another widely applied traffic calming measure, are roads with special characteristics, which allow safe walking. Vehicles, although allowed in, move with very low travel speeds (up to 30 km/h) and the priority is yield to pedestrians. Such measures are constructed in one-way roads, as well as in roads of two directions, and the respective road widths are 3 m and 5 m. The above-mentioned road types impend the free flow of vehicles; consequently traffic volumes are significantly reduced. However, they do not cause feelings of annoyance to the drivers, as it happens with speed humps. The construction and maintenance costs are much higher. Figure 13 presents the ground plan of a road with mixed circulation of vehicles and pedestrians (woonerf). The possibility of parking is very limited, while the presence of trees is intense. In this way, the aesthetics of the local environment is upgraded and the green in the urban regions is increased. Finally, as indicated in Figure 13, vehicles are not allowed to move straight ahead, but are constrained to follow an “S” manoeuvre. Figure 13: Ground plan of a Consequently their speed does not exceed the woonerf. relevant speed limit that is in Source: Magee, 1998 effect for such roads, i.e. 30 km/h. Woonerfs are constructed in most developed counties together with speed humps (or bumps), roundabouts, traffic circles, raised intersections, median barriers or islands, curb extensions and chokers, chicanes or street closures.

2.2 Description of areas where traffic calming measures were implemented

In Athens, the capital of Greece, a limited number of traffic calming measures has been constructed. The Municipality of Neo Psychiko is the only area in the Greater Athens Area, which inaugurated an extensive road traffic calming programme at the beginning of 1990’s in an attempt to improve road safety in this area. A wide range of traffic calming measures was carefully implemented, according to technical specifications. These measures mainly included speed humps and woonerfs and were basically implemented between the years 1991 and 1999. (Municipality of Neo Psychiko, 2001). Neo Psychiko is the area of investigation of the impact of Low Cost Traffic Engineering Measures on road safety in urban areas and the methodology used is the “before and after accidents analysis with large control group”. The control group chosen consists of the neighbouring Municipalities of Holargos and Agia Paraskevi in the Athens Greater Area. It is important to mention that in this research only streets with one direction and one lane are examined, as in this type of streets traffic calming measures were primarily implemented.

Page 101 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

3 Target group

The inhabitants of the Municipality of Neo Psychiko mainly benefit from the implementation of the traffic calming measures in the area. Especially the vulnerable road users groups (pedestrians, children, two-wheelers, pedal cyclists) are considered as the target group, but the reduction of road accidents also concerns the drivers and passengers circulating in the area. 4 Assessment method

4.1 General

Cost-benefit analysis (CBA) is the financial tool used for the economic appraisal of the installation of speed humps and woonerfs in the Municipality of Neo Psychiko. Generally, CBA provides a logical framework for evaluating alternative courses of action when a number of factors are highly conjectural in nature. Essentially, it takes into account all the factors that influence either the benefits or the cost of a project, even if monetary value can not be easily assigned [SMITH, 1998]. For the purpose of this research, the main benefit (safety effect) considered in the calculations is the number of prevented accidents in the area, after the implementation of traffic calming measures. Social and environmental effects for the residents of the area are not taken into account in this study, as it is difficult to be quantified and moreover, their benefits are not essential comparing to the accident reduction. However, the time lost (for the road users) due to the reduction of travel speed should be incorporated into the benefits calculation.

4.2 Estimation of safety effect

Although there is a wide variety of methodologies used for the examination of road safety in an area, for the estimation of the safety effect in the Municipality of Neo Psychiko, deriving from the implementation of speed humps and woonerfs in the area, the “before and after methodology with large control group” was considered. This is the methodology with the highest degree of accuracy, as the size of control group is quite large and moreover, when there is a sufficient number of years “before” and “after” the implementation of traffic calming measures (as it is in this case study), the phenomenon of the regression to the mean is eliminated, making the “before and after methodology with large control group” the most appropriate and reliable methodology for the estimation of the potential safety effect. The effects observed in the treated area and the control group area, are weighted by means of Odds-ratio of the total number of road accidents in “before” and “after” treatment period. This results to the estimated effect:

Page 102 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

Estimated effect (θi ) = [Xa/Xm]/[Ca/Cb] where

Xa - the number of road accidents observed at the treatment area in the “after“ period

Xm - the number of road accidents observed at the treatment area in the “before“ period

Ca - the number of road accidents observed at the control group area in the “after“ period

Cb - the number of road accidents observed at the control group area in the “before“ period

The statistical weight of the estimate is: 1 w = i 1 1 1 1 + + + Ai B i C i D i Where A, B, C, D are the four numbers of the odds-ratio calculation. The weighted mean effect is :

∑ wi ln(θi ) Weighted mean effect(WME) = exp( i ) ∑ wi i with 95% confidence interval for the weighed effect estimated as follows:

     z α z α     1−  WME exp 2 ,WME exp 2    w   w    ∑ i   ∑ i    i   i 

The applicable value of the safety effect, i.e. the best estimate of accident reduction associated with the treatment (in percents), is calculated as (1-WME)*100. The control group should include large areas with similar characteristics to the area considered, where traffic calming measures were not implemented. The Municipalities of Holargos and Agia Paraskevi in the Athens Greater Area present similar road network, population density, land use and traffic volumes characteristics with the Municipality of Neo Psychiko (area considered), as indicated in Table 42 and were therefore chosen as the large comparison group (Georgopoulou, 2002).

Page 103 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

Table 42: Road network, land use and other characteristics for the area considered and the control group Municipality Characteristics Neo Psychiko Agia Paraskevi Holargos Road network characteristics Area’s extent 1,200,000 m2 7,000,000 m2 2,735,000 m2 Population 16,000 87,500 39,000 Density 130 res/acre 125 res/acre 142 res/acre Number of blocks 174 514 250 Average surface of each block 6.90 m2 13.62 m2 10.94 m2 Road network length 19,000 m 120,000 m 42,000 m Basic road network length 3,700 m 25,000 m 7,000 m Secondary road network length 15,300 m 95,000 m 35,000 m Road surface percentage 12.63% 13.79% 12.03% Number of streets 75 288 95 Number of one direction streets 67 260 85 Number of two directions streets 8 28 10 One direction streets 89.33% 90.28% 89.47% percentage Two directions streets 10.67% 9.72% 10.53% percentage Number of secondary streets 8 15 11 Land use and other characteristics Over-regional business land use 16.21% 14% 12% Regional business land use 1.44% 1.02% 1.2% Land for cultural events etc. 3.8% 4% 3.3% Education + Sports 4.93% 4.2% 3.5% Residence 73.62% 76.78% 72% Monthly family average income 1350 € 1100 € 1100 € 49% with 45% with public 55% with public public transport transport Number of trips (to the centre of transport Athens) 51% with 55% with private private 45% with private vehicles vehicles vehicles Average residents’ vehicle 300 – 350 350 – 400 350 – 400 property (vehic/1000 resid.)

Page 104 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

5 Assessment Quantification

5.1 Traffic calming measures implementation cost

The total cost for the implementation of traffic calming measures in the Municipality of Neo Psychiko can be distinguished into implementation costs for speed humps and implementation costs for woonerfs. The cost of speed humps includes the designing and construction/installation costs, depending on the type of material used (asphalt or plastic) as well as the respective road markings. In the case of Neo Psychiko, 49 speed humps were installed in 21 one-lane, one-direction roads and the total cost was 117.390€ (1998 prices). The implementation cost of woonerfs is considerably higher than the respective of speed humps, as it concerns larger areas and includes the design cost, cost for the configuration and pavement of the respective areas, cost for hydraulic works, electrical works and sewage pipelines installation. In the case of Neo Psychiko, a total area of 100,000 m2 in 40 local roads was transformed into woonerfs between 1991-1999. According to the data provided by the technical department of the Municipality of Neo Psychiko, 4,402,054 € (at 1998 prices) was the total cost for the implementation of woonerfs, which is considered quite high. Generally, increased construction cost is a particularity of the Greek tendering system. The above-mentioned implementation costs are shown in Table 43. Table 43: Traffic calming measures implementation cost

Traffic calming measures Amount Cost Speed humps 49 units € 111,518 Woonerfs 100,000 m2 € 3,081,438 Total Implementation Cost € 3,192,956 *1998 prices

5.2 Traffic calming measures benefits

In the framework of this research, the benefits examined exclusively concern safety benefits deriving from the reduction of all injury accidents in the examined area, as no significant social or environmental costs were expected from the implementation of speed humps and woonerfs in the Municipality of Neo Psychiko. The available results of previous research allowed for the direct calculation of the number of accidents prevented by the measures, as described in detail in the following sections.

5.2.1 Number of accidents prevented

After the resemblance of the area examined and the control group was proved, the “before and after” methodology was applied to examine the statistical significance of the reduction of road accidents in the area where traffic calming measures were implemented. The evaluation of the safety effect, which in this case study is the number of all injury accidents prevented, is based on the Test X2. The number of accidents occurring in the area examined is compared with the accidents occurring in the control group. More specifically, X and Ψ represent, respectively, the total number of accidents that occurred in the period before and after the implementation of the measures in the area considered.

Page 105 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

Similarly, XE and ΨE represent, respectively, the total number of accidents that occurred in the control group area, where traffic calming measures were not implemented. The Test X2 gives that:

(Ψ - ΧΑ)2 Χ2 = (1) (Χ + Ψ)Α

Ψ where Α = Ε (2) ΧΕ

2 2 Then, the estimated X value is compared with the X α value for a given probability standard α and for n = 1 freedom standard (n = k–1, where k = 2 are the observations, one before and one after the implementation of the measures), as they are given in relevant tables. 2 2 When the estimated X value is higher than the Χ α (for a predetermined probability standard α), the reduction in the number of accidents is considered statistically significant and in all likelihood is attributed to the implementation of speed humps and woonerfs. The pre-determined probability standard (α) used in this research is 95%, which can be considered as conservative. The total number of accidents occurred in one direction: one-lane streets in the area of Neo Psychiko during the years 1985-1990 and during the years 1994-1999 are 36 and 33, respectively. Similarly, the total number of accidents recorded in the control group is 101 and 149, respectively. According to the previous symbolism, X = 36, Ψ = 33, ΧΕ = 101 and ΨΕ = 149, as indicated in Table 44.

Table 44: number of accidents “before” and “after” in one direction - one lane streets

Area Time period Area examined Control group (Neo Psychiko) (Xolargos and Agia Paraskevi)

Before (1985-1990) Χ = 36 ΧΕ = 101

After (1994-1999) Ψ = 33 ΨΕ = 149

Proportion -8.3% 47.5%

After applying equation (1), it is estimated that: X2 = 3.972 > 3.84 (X2 value for 95% probability standard), so that a statistical significant reduction in the total number of accidents is noticed. A reduction of 8,3% in the total number of accidents was observed in the area considered, while an increase of 47,5% was recorded in the region of control group. In Table 5 the mean value of the estimated safety effect and the confidence interval for this value are presented.

Page 106 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

Table 45: Safety effect of speed humps and woonerfs estimated for Neo Psychiko Treatment type Estimated effect (WME) WME confidence interval Speed humps and woonerfs in the 0.621 (0.363, 1.061 ) Municipality of Neo Psychiko

The average safety effect of speed humps and woonerfs implementation in Neo Psychiko is a 38% reduction in of the total number of road accidents, thus, 14 accidents were prevented by the presence of these traffic calming measures, as no other road safety measure occurred in the area at the same period.

5.2.2 Accident cost

The estimation of average accident costs was carried out on the basis of a recent study on accidents cost in Greece [LIAKOPOULOS, 2002]. This study concerned the estimation of the costs of various components of accident costs (material damage costs, generalized costs, human costs) for fatal accidents, injury accidents and material damage accidents, including: • Material damage costs • Police costs • Fire brigade costs • Insurance companies costs • Court costs • Lost production output • Pain and grief • Rehabilitation costs • Hospital treatment costs • First aid and transportation costs The various costs were calculated by means of an exhaustive data collection process addressed to various organizations (National Statistical Service of Greece, National Police, Fire Service of Greece, Emergency Medical Service of Greece, hospitals, courts, insurance companies etc.). Additional parameters were adopted on the basis of estimations from experts in each field, as well as the existing international literature. It should be noted, however, that the above study did not adequately account for the human cost component, as the pain and grief parameters (reported in the Courts) are not sufficiently representative of the human cost. On that purpose, a separate investigation for human costs in Greece was carried out in the framework of the present research. In particular, human costs was estimated according to the following formula: VoSL = (NAEIS) / (LSE) Where: VoSL: Value of Statistical Life NAEIS: National Annual Expenditure on Improving Safety LSE: Expected Lives Saved from this Expenditure Annually

Page 107 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

In particular, the calculations included parameters such as the percentage of the family annual income that each person is willing to pay in his/her entire life in order to reduce the probability of accident involvement of himself/herself or of any family person by 50%, the average members per family in Greece, the proportion of families with an economically active member, the average family annual income in Greece, the national population, the life expectancy in Greece and the current and new accident risk. With regards to the percentage of the family annual income that each person is willing to pay in his/her entire life in order to reduce the probability of accident involvement by 50%, the results of a recent "willingness-to-pay" survey in Greece were used [AGGELOUSI, KANELLOPOULOU, 2002]. In this survey, drivers were asked to state the percentage of annual income they are willing to pay to reduce the probability of a fatal accident, an injury accident and a material damage accident involvement by 50%. Furthermore, they were also asked to rate various types of accidents and injuries, in order to identify their perception on injury severity. On the basis of the results, in the present research the value corresponding to injury accidents is considered to adequately represent serious injury accidents, whereas the value for material damage accidents is considered to adequately represent both minor injury and material damage accidents. On the basis of the above, the human cost of accidents in Greece was estimated as follows: VoSL = 612,140.72 €/person for fatal accidents VoSL = 467,703.02 €/person for serious injury accidents VoSL = 206,339.57 €/person for slight injury and material damage accidents It should also be underlined that the calculations concern prices for 1999. In order to calculate the average accident cost in Greece, the costs of fatal and injury accidents were weighted in relation to the average distribution of accident casualties per casualty severity in urban areas in Greece. In the following Table 46, parameters concerning accident costs in Greece are summarized on the basis of the previous research used and the additional calculations carried out. Table 46: Calculation of average accident cost in Greece (1999 prices)

Cost of Accidents with: Killed Seriously Injured Slightly Injured Material Damage cost (€) 28,769.42 18,174.91 13,904.19 Generalised cost (€) 442,466.54 23,906.66 6,960.30 Human cost (€) 612,140.72 467,703.02 206,339.57 Total cost (€) 1,083,376.68 509,784.59 227,204.06 Proportion of casualties in urban areas 3.70% 9.11% 87.19% Average accident cost 284.666,63 €

5.2.3 Estimation of cost for time lost

The implementation of traffic calming measures in an area results to reduced travel speeds (a reduction of 8 km/h – 15 km/h is usually observed). The time lost (for the road users) due to this speed reduction could also be incorporated into the benefits calculation as a negative effect and its value is estimated according to the following equation: T = D * Q * V * P

Page 108 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

Where T: the value of time lost due to delays resulting traffic calming measures implementation D: average delay per vehicle Q: average daily traffic volume in the area considered V: average value of time (hourly) per vehicle P: period The average delay per vehicle (time lost due to implementation of speed humps and woonerfs) when circulating in the area of Neo Psychiko is approximately 60 seconds. This estimation is based on field measurements, which took place in the area considered. The average daily traffic volume in the Municipality of Neo Psychiko was 8,680 vehicles. The hourly cost of the delay of an average vehicle is 4.5 €/hour (1999). This calculation takes into account the average value of time per person (hourly) for 1999, which is 3 €, as well as the average vehicle occupancy, which is 1.6 € [ATTIKO METRO, 1997]. Finally, the examined period is the number of working days over a year (260 days). Consequently, the value of time lost in the area considered due to traffic calming measures implementation is: T = 60 sec/vehicle * 8,680 vehicles/day * 4.5 €/hour * 260 days * 1/3,600 hours = 180,544 € (1999 prices).

6 Assessment Results

The cost-benefit ratio calculation follows the identification and quantification of the costs related to the implementation of traffic calming measures and their benefits, described in the previous sections. An accumulated discount factor was applied to the implementation cost calculation on the basis of an interest rate of 4% [National Statistical Service of Greece, 2003]. Two scenarios are developed, according to the calculation of the value of benefits. In the first scenario, the value of benefits derives only from the number of accidents prevented in the area (scenario 1) and in the second scenario the yearly value of time lost in the area due to traffic calming measures implementation is also considered (scenario 2). On that purpose two ratios are calculated: Table 47: Calculation of the cost-benefit ratio Scenario 1 Scenario 2 Safety benefits only Including time lost Present value of benefits Number of accidents prevented 14 14 Average accident cost - 1999 (€) 284,666.63 284,666.63 Accumulated discount factor 1.0 1.0 Value of time lost - 1999 (€) - 180.544 Total (€) 3,985,332.82 3,804,788.82 Present value of costs Implementation cost - 1998 (€) 3,192,956.71 3,192,956.71 Accumulated discount factor 1.04 1.04 Implementation cost - 1999 (€) 3,320,674.98 3,320,674.98 Cost-benefit ratio 1.2:1 1.14:1

Page 109 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

The yielded cost-benefit ratio indicated in the above Table 47 proves that the implementation of speed humps and woonerfs in a broad local area can be cost-effective.

7 Decision-Making Process

The results of this research were presented to Head Officers of the Technical Department of Neo Psychiko. As these decision-makers are mainly civil engineers, they are familiar with efficiency assessment in terms of cost-benefit analyses and they responded positively towards this work from the first stages, contributed with data and other available information and were very helpful in dealing with lack of data when necessary. Furthermore, decision-makers were very interested in the results. The cost-benefit ratios, although not very high, were received as a confirmation of the important role of the local authorities in road safety improvement of urban areas and a validation of their systematic efforts to contribute in the reduction of road accidents and casualties in their municipality. Even though the implementation cost of the traffic calming measures are considered relatively increased, they believe that the reduction in accidents and the respective lives that can be saved are worth every possible effort. Consequently, they intend to continue the implementation of similar road safety measures and they would like to communicate these results to the residents of Neo Psychik, to the press, as well as to other municipalities so they can also benefit. They also added that if the results were negative or even less encouraging, they would try to identify the more cost-effective cases among the results and focus their efforts accordingly, or consider alternative and more efficient road safety related activities. Decision-makers also expressed a high interest for more analyses and results, concerning implementation of other traffic calming measures, in more road types than one-lane - one direction, or the results concerning specific types of road users (e.g. pedestrians, two- wheelers and elderly people). They also underlined that these results would have been even more useful if they were available at earlier stages of the implementation of the speed humps and woonerfs and they expressed their strong willingness to mutually co-operate with any responsible authorities in order to further improve the road safety of their area.

8 Implementation barriers

As far as the implementation of traffic calming measures is concerned, the basic barrier refers to the reactions from all drivers using the streets where the speed humps and woonerfs were installed. The reduced travel speeds, as well as the negative impact of such measures on the suspension system of the vehicles and the relevant annoyance to the drivers, lead very often to complaints. Some of these road users are residents of the area and some others are just passing through. Moreover, the elaboration of guidelines and standards for the construction and maintenance of the road network in Greece (even at the local level) is a task for the Ministry of Public Works. In the case of traffic engineering measures, such guidelines and technical specifications do not exist and consequently their development by the technical department of Neo Psychiko and the relevant governmental authorities resulted in delays during the implementation phase. These parameters were the main difficulties encountered during the early implementation period.

Page 110 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

Additionally, it should be emphasized that such a research should be complemented with other studies concerning other disadvantages stemming from the implementation of traffic calming measures. More specifically, it is very essential that the negative impact of these measures on the traffic flow of vehicles should be examined. Speed humps or woonerfs in urban streets result in reduced car speeds, which in turn, affect adversely the traffic flow of streets and causes the undesirable “immigration” of accidents to adjacent streets [FRANTZESKAKIS, GOLIAS, 1994]. Furthermore, the possible negative impacts of speed humps on the suspension system of cars should be considered while calculating the value of benefits. The extent of this damage is highly dependent on the size and geometrical characteristics of those devices, as well as the speed of passing cars through them, and it comprises one of the most controversial aspects related to the implementation of traffic calming measures in urban areas. The lack of appropriate data for cost-benefit evaluation purposes and the fact that neither local, nor governmental authorities have used any economic evaluation tools so far to demonstrate the correctness of decision-making, were overcome by means of interviews with transport engineers from the technical department of the Municipality of Neo Psychiko, who were also actively involved in both the decision-making process and the monitoring of the traffic calming measures influence on road accident reduction. Additionally, existing research in Greece was further used to yield the necessary parameters for the computation of cost/benefit ratios.

9 Conclusion / Discussion

There is a certain correlation between low cost traffic engineering measures in urban areas and the respective number of road accidents. International experience in many developed countries has shown that several of the traffic calming measures (speed humps, woonerfs, raised intersections, road narrowing, etc.) are deemed to be the most efficient measures towards tackling one of the most significant problems that communities face nowadays: urban road accidents. In Greece such measures were implemented only in few municipalities and in most cases the implementation was either incomplete or not well prepared. A first approach for reliable and comprehensive evaluation of the effectiveness of those measures in reducing accidents, speeds or casualties is attempted through this research, as no evaluation studies have been undertaken so far in Greece. The present research revealed very limited use of assessment methods in the overall decision-making process in Greece. Only a small number of cost-effectiveness studies on road safety measures in general were conducted systematically by independent institutions and organisations. These occasional research initiatives provide some insight on the existing activities, but scarcely lead to interesting conclusions and thus are not usually transferred to policy-makers. In this study, the cost-benefit analysis was applied to an urban area (a municipality) in order to evaluate the economic effectiveness of certain traffic calming measures (speed humps and woonerfs). The safety effect (reduction of number of road accidents in the area) deriving from the implementation of such measures was calculated and statistically evaluated by applying the “before and after” methodology with large control groups. Monetary value was assigned to this safety effect by calculating the average accident cost.

Page 111 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

Special consideration was given to the estimation of human costs, which is an ambiguous component of the total accident cost for different casualty types. Data on the measures’ implementation costs were provided by the technical department of Neo Psychiko and the cost-benefit ratio was calculated for two different scenarios, according to the calculation of the benefits’ value. However, the incorporation of the time lost in the value of benefits (scenario 2) did not really affect the result, as according to scenario 1 the estimated ratio was 1:1.8, whereas in scenario 2 the ratio was calculated as 1:1.7. In both cases the ratio shows that traffic calming measures’ implementation is cost- effective. The fact that the cost-benefit ratio is not very high could be attributed to the high implementation cost of the traffic calming measures, a particularity of the project tendering system in the Greek construction sector. Finally, it was worth mentioning that the absence of national and co-ordinated road safety programmes aiming at accident reduction can be overcome by the successful implementation of several road safety actions at the local level, like the traffic calming measures in urban areas. Generally, close cooperation of governmental and regional or local authorities can be very effective in road accident improvement at the local level. The cost-benefit analysis indicates that traffic calming measures could be a useful tool in the hands of decision-makers when considering road accident reductions in urban areas, although the implementation cost is high in several cases and there are several complaints from the road users concerning the reduced travel speeds. However, it is society who has to choose between speed and safety.

Page 112 IMPLEMENTATION OF LOW COST TRAFFIC ENGINEERING MEASURES AT MUNICIPALITY LEVEL

REFERENCES

AGGELOUSSI, K., KANELLOPOULOU, A., (2002): Estimation of the human cost of road accidents and drivers' sensitivity towards accident risk - A willingness-to-pay technique and a stated-preference technique, Diploma Thesis, NTUA, School of Civil Engineering, Department of Transportation Planning and Engineering, Athens. ATTIKO METRO SA. (1997): Land use characteristics and socio-economical parameters. (03). P.S: Land use characteristics and density of the Greater Area of Athens. DEPARTMENT OF TRANSPORTATION PLANNING AND ENGINEERING, (2004): Accident risk investigation of categories of drivers with high accident involvement - second report, Ministry of Transportation and Communication. ENGEL U, THOMSEN L., (1992): Safety effects of speed reducing measures in Danish Residential Areas, Accident Analysis & Prevention. Vol. 24, No 1, pp. 17 -28. FRANTZESKAKIS G., GOLIAS G., (1994): Road Safety, Papasotiriou Publications. GEORGOPOULOU X., (2002): Investigation of Low Cost Engineering Measures’ Impact on road safety in urban areas, Diploma Thesis, NTUA, School of Civil Engineering, Department of Transportation Planning and Engineering, Athens. JACKSONVILLE FLORIDA CITY, (2002): Neighbourhood Traffic Calming Manual, Traffic Engineering Division. KANELLAIDIS G. et al., (1999): Attitude of Greek drivers towards road safety, Transportation Quarterly. KANELLAIDIS G, et al., (1995): A survey of drivers’ attitude towards speed limit violations. Journal of safety Research. KAPICA C., (2001): Pilot Study Report on Speed humps, Columbia Avenue Hartsdale, New York. (www.town.greenburgh.ny.us/Speedhump.pdf) LIAKOPOULOS D., (2002): Development of a model for the estimation of the economic benefits from accident reduction in Greece, Diploma Thesis, NTUA, School of Civil Engineering, Department of Transportation Planning and Engineering, Athens. MUNICIPALITY OF NEO PSYCHIKO, (2001): Regional Development Programme, Technical Division of the Municipality of Neo Psychiko. NATIONAL STATISTICAL SERVICE OF GREECE, (2003): "Greece in figures, Official Publication of the National Statistical Service of Greece, Athens (www.statistics.gr). SMITH N., (1998): Engineering Project Management, Blackwell Publication. ZAIDEL D. et al., (1992): The Use of road Humps for Moderating Speeds on Urban Streets, Accident Analysis & Prevention, Vol. 24, No 1, pp. 45 - 56.

Page 113 CASE F1: Grade-separation at railroad crossings

ROSEBUD WP4 - CASE F REPORT

GRADE-SEPARATION AT RAILROAD CROSSINGS

BY MARKO NOKKALA,

VTT BUILDING AND TRANSPORT, FINLAND TABLE OF CONTENTS

1 PROBLEM TO SOLVE ...... 117 2 DESCRIPTION OF MEASURE...... 118 3 TARGET GROUP ...... 118 4 ASSESSMENT METHOD...... 118 5 ASSESSMENT QUANTIFICATION...... 121 6 ASSESSMENT RESULTS...... 124 7 DECISION MAKING PROCESS...... 124 8 ROLE OF BARRIERS ...... 125 9 DISCUSSION...... 125 GRADE-SEPERATION AT ROADRAIL CROSSINGS CASE OVERVIEW

Measure: Grade-separation of at-grade rail-road crossings Problem to solve: Train-vehicle collisions at the crossing (and vehicle delays due to crossing's closures) Target Group: Train-vehicle accidents Targets: Diminishing accidents and traffic delays Initiator VR, the Finnish national Railway Authority (also linked to National Road Administration) Decision-makers Ministry of Transport and Telecommunications (on the level of targeting specific measures), Road Authorities, Railway Authorities Costs: Investments in grade-separation construction; by the Railway Authority Benefits: The benefits are accident savings and a reduction in traffic delays. Driving public will benefit. Cost/Benefit-Ratio: For a rural crossing the CBA ratio is 0.65; for the urban crossing the ratio is 0.25.

Page 116 GRADE-SEPERATION AT ROADRAIL CROSSINGS 1 Problem to solve

The large majority of rail-road crossings in Finland, like in any country, are level (at-grade) crossings. In general, at-grade rail-road crossings are associated with economic losses due to vehicle delays and train-vehicle collisions. In the Finnish context, level crossings have been considered the cost-effective measure to construct crossings, due to the fact that traffic volumes at points of crossings have been small. However, in the late 1990s the awareness of need to upgrade existing crossings, either through increased safety measures or construction of grade-separation crossing increased rapidly, following some severe accidents at the crossings. To illustrate the situation in numeric figures, over the past years, 1999-2003, Table 48 lists the statistics of deadly and severely injured accidents. As the statistics show, there has been an observable increase in deaths per 1 million passengers in 2000 and 2001. In 2003 a total of 17 persons were killed in rail accidents, with the following breakdown: • Level crossings with warning signals: 2 • Level crossings without warning signals:4 • Other, non specified: 11

Table 48: Accident statistics in Finnish rail, 1999-2003

TYPE OF ACCIDENT 1999 2000 2OO1 2002 2003 Death and seriously injured 0,72 1,00 1,03 0,57 0,71 per 1 million passenger kms Accident cases per 1 million 2,10 2,01 2,18 1,72 2,00 passenger kms Deaths per 1 million 0,02 0,04 0,04 passengers Seriously injured per 1 million 0,11 0,05 0,09 0,02 0,02 passengers

There has been a significant research program of the VTT Building and Transport to study the needs to upgrade the out-of-date crossings facilities in Finland. It has been found that a significant part of locations with accident occurrences are at-grade crossings which are equipped with automatic safety gates (i.e. have the highest form of safety protection for at- grade crossings), but in some cases there have been no safety gates due to the low volume of crossings. Due to high train frequencies and significant road traffic volumes at some crossings, the economic losses because of vehicle delays might be high. Therefore, the question was to point out the sites where a grade-separation is warranted. The process of grade-separation is expensive. It is therefore essential to provide a systematic approach for decision-makers that will lead to a considered decision on the benefits and costs associated with grade-separation. However, a detailed investigation of a specific crossing is time-consuming and costly (Tustin et al. 1986; Taggart et al. 1987), and cannot be reasonably performed for a large number of sites. Thus, at the initial stage, screening tools are required that will assist in

Page 117 GRADE-SEPERATION AT ROADRAIL CROSSINGS choosing, from the whole set of locations (i.e. from the whole railway network), those warranting further consideration. There is a set of screening tools for crossings’ consideration for grade-separation developed by the study Gitelman, Hakkert (2001) in Israel. The tools consist of a safety model, a formula for estimating the economic loss due to vehicle delays at a crossing and a qualification criterion. The tools are based on economic principles, comparing the economic loss due to an at-grade crossing with the average cost of grade-separation. In this study we will combine the information of delays from the Israel model with other data and methods used in Finnish standard appraisal. In this report present a cost-benefit analysis (CBA) of grade-separation of two representative rail-road crossings, from rural and urban settings in Finland.

2 Description of measure

A grade-separation of a rail-road crossing means building a bridge or a tunnel instead of existing at-grade crossing. A grade-separation eliminates existing railway-road crossing and consequently, removes the problem of train-vehicle collisions at the site considered. Besides, the grade-separation considerably diminishes the amount of road traffic delays at the site which previously stemmed from the crossing’s closures due to trains’ movements. A grade-separation is usually considered for rail-road crossings which are already protected by automatic gates and where the frequency of accidents due to, for instance, exceptional circumstances, is high.

3 Target Group

The target accident group are train-vehicle collisions at level crossings. The project aimed at developing screening tools for selecting crossings with high potential for grade-separation, i.e. those crossings where the costs of vehicle delays and safety problems associated with the at-grade crossings are sufficiently high in order to justify building a grade-separation. Such tools are needed for decision-makers as they both stimulate an objective policy and a systematic approach to the issue, and define a priority for grade separation at crossings. The tools are applied to perform a CBA of a grade-separation of two typical types of crossings.

4 Assessment method

4.1 Assessment tools developed

The economic losses associated with the current situation, i.e. at-grade crossing, stem from two main factors: vehicle delays and safety problems. These losses represent the economic benefits which can be attained due to eliminating at-grade crossing. The CBA should compare these potential benefits with the costs of building a grade-separation. The assessment tools developed for estimating potential benefits from a grade-separation include (Gitelman, Hakkert, 2001):

Page 118 GRADE-SEPERATION AT ROADRAIL CROSSINGS

1. An accident prediction model, which, along with accident costs, supplies a basis for evaluating losses due to safety problems at level crossings. 2. A model for evaluating economic losses due to vehicle delays at any crossing, based on its parameters. 3. A quantitative criterion for grade separation which combines the results of both models. The principles, designed for the Israel, also apply for the Finnish case, despite using different tools for evaluation and modelling in Finland. In Finland, the common tool for economic appraisal of transport projects is socio-economic profitability calculations, which are based on calculating the vehicle costs, time savings (or losses), accident changes, pollution and noise costs of the investment project. This method is applied in this study to ensure compatibility with other appraisals in Finland.

4.1.1 Evaluating economic losses due to safety problems

Safety concerns at level crossings frequently provide the main reason for grade-separation (Europe@ 1998; US GAO 1995). The evaluation of the safety factor needs two inputs: an estimate of the expected number of accidents per crossing and the cost of an average crossing accident. The expected number of accidents per crossing represents the annual number of accidents which will be saved due to grade-separation, whereas their costs demonstrate the economic value of safety benefits expected. In Finland rail statistics collect data on annual accidents and for each type of accident there is a specified value to be used in estimating the monetary loss resulting from the accident. In Finland both the Railway and Road authorities have systems to collect accident data with location-specified. This means that for each of the crossings it is possible to collect history data on accidents, traffic volumes and other relevant information. There is a model called TARVA in use in Finland to estimate the accidents data. TARVA can be used to calculate the probabilities of accidents on the specified location on the roads network. However, the calculation of accidents and their prevention in the selected crossings proved difficult as there were only minor accidents on the locations. The following Table 49 summarises the unit values used for various types of accidents. The unit values are confirmed by the Finnish Ministry of Transport and Telecommunications and the figures were last revised in 2000. Particularly the compensation for severe accidents has risen over time, reflecting changes in the method to shift towards willingness-to-pay method.

Table 49. Unit values for accidents. (Ministry of Transport and Telecommunications 2003). Accident with severe injury damages, € 386,832.00 Accident leading to death, € 2,430,316.00 Average value of the accident, € 84,094.00

4.1.2 Evaluating economic losses due to vehicle delays

We utilise the evidence from Israel to supplement the Finnish evaluation method of transport project to estimate the vehicle delays. This is useful since there are no accurate Finnish values available for this type of delay (which is often considered too small an item

Page 119 GRADE-SEPERATION AT ROADRAIL CROSSINGS to be accurately measured). In Israel, a sample of 20 crossings was selected for detailed measurement. The crossings were selected from among the busier lines of the rail network. At each crossing, the parameters measured were as follows: vehicular traffic volumes; closure times and queue release times; and vehicle speeds. For traffic volumes, hourly distribution was attained, along with the traffic subdivision into three vehicle classes: cars, trucks and buses. Prior to the evaluation, the hours were divided into three time intervals, according to the traffic volumes observed: peak, low (night) and intermediate volumes. The closure times (of the automatic gates) were estimated for three types of trains: passenger trains, freight trains and operational trains. Vehicle speeds were measured at a “free” distance from the crossing and emmediately before the crossing. Besides, for each crossing, the number of train transitions per each hour was calculated based on the railway line operative time-table. The analysis revealed that the closure times depend on the train type, train speed and the vicinity of station, whereas for the times for queue release no clear dependence was seen between this parameter and the average traffic volume or closure time (Gitelman, Hakkert, 2001). The annual cost of vehicular delays at a crossing was estimated from:

D = 260⋅[N ⋅ d1 + (V − N) ⋅ d2 ] (1) where D=annual cost of vehicular delays, euros (for 260 working days a year), V = vehicular daily traffic volume, vehicles, N = number of vehicles stopped at the crossing per day, d1 = average cost of a vehicle’s stopping at the crossing, euros, d2 = average cost of a vehicle’ slowing down at the crossing, euros. The economic losses sustained from traffic delays ensue from additional consumption of fuel and other vehicle expenses and from the time lost to vehicle occupants because of “velocity cycles” when passing the crossing. Estimating d1 and d2 in the above formula, the losses due to different vehicle and train types at a specific crossing were weighted, in accordance with the shares of these types in daily vehicle/ train traffic at this site. The detailed calculation of vehicle delay costs at a specific crossing consists of various and multiple data considerations. Hence, for a rapid screening of sites, an approximate formula was developed which allows estimation, based on the crossing’s parameters and without prolonged calculations. The model fitting was performed by means of the SAS multiple linear regression module, where the parameters and estimates of the sample crossings served as a database. The approximate formula recommended for application was (Gitelman, Hakkert, 2001): Y/3.79 = -0.656044 + 0.000108*V + 0.0023038*Trains + 0.094042*Slowdown (3) where Y/3.79 = annual economic loss due to vehicle delays at a crossing, million euros (where 3.79 is the exchange rate of NIS/euro), V – daily traffic volume (vehicles), Trains – daily number of trains (trains), Slowdown – average vehicle speed reduction due to a crossing (km/h). Page 120 GRADE-SEPERATION AT ROADRAIL CROSSINGS

4.2 Considering cost of the measure

Summing up the costs of vehicle delays and of safety problems provides a value of annual economic loss at the level crossing, i.e. the magnitude of economic benefits, which could be attained due to a grade-separation. This value should be compared with the construction costs. Urban grade separations tend to be larger projects and have higher costs than rural crossings, with a range of € 5.0 million to urban and € 2.9 million for rural crossings, at 2000 prices, would be a reasonable average for the construction of a grade separation at a Finnish crossing. Considering the net present value of the construction costs, with a 5 per cent discount rate used in the Finnish project appraisal and a 20-year project life18, supplies a value of benefits (or annual economic loss at a level crossing) that would justify a grade-separation. We note that the cost of the project is considerably higher in the rural context when calculated per crossing vehicle as opposed to urban crossings. In Finland the decision-making on grade-separation appears to be non-linked to the economic benefits of the upgrading, but rather on the comfort and safety of travel, expressed in non-monetary terms. This is evident from the fact that the costs tend to be reasonably high in the rural context, which is itself a factor hindering the developments but also leads to decision-making where crossings are built independent of their costs.

5 Assessment Quantification

5.1 General

In this section we consider a CBA of grade-separation of two different types of crossings. To note, a cost-benefit and not cost-effectiveness analysis was chosen, due to following reasons: 2. Standard project appraisal in Finland on transport sector is based on cost-benefit, not cost-effectiveness analysis. 3. Monetary valuations of all benefits and costs should be applied (inter alia, to justify the implementation of the measure). The main data elements to be provided for the CBA performance are (WP3, 2004): • A definition of unit of implementation for the measure; • An estimate of the number of accidents are expected to prevent per unit implemented of the measure, through: identification of target accidents, estimate of the number of target accidents expected to occur per year, estimate of the safety effect of the measure on target accidents; • Accident costs; • Other monetary values depending on the effects considered; • An estimate of the costs of implementing the measure; • The economic frame for the evaluation (length of service life, interest rate).

18 According to Finnish recommendations for economic evaluation of transport projects Page 121 GRADE-SEPERATION AT ROADRAIL CROSSINGS

In our case of grade separation of at-grade crossings, the above data elements will be as follows: • The unit of implementation is one at-grade crossing; • Target accidents are all train-vehicle accidents at the at-grade crossings. The number of target accidents expected to occur per year can be estimated using a prediction model, TARVA. The safety effect of the measure is 100% reduction in target accidents, as a grade-separation implies the elimination of all train-vehicle collisions. Thus, in this case, the number of accidents are expected to prevent following implementation of the treatment is equal to the number of target accidents which are expected to occur at the site, prior to implementation of the treatment. • Accident costs – see Section 4.1.1; • Other monetary values include costs of travel time and vehicle operating costs; they can be estimated using formulae from Section 4.1.2; • The average cost of implementing the measure – see Section 4.2; • The economic frame for the evaluation: 20-year project life, with 5% discount rate. The CBA is performed for two at -grade crossings: one in the rural context (Outinen) and one in the urban context (Hennala).

5.2 CBA of a rural crossing

Crossing at Outinen is a rural rail-road crossing, which is situated on the Kouvola- Pieksämäki railroad section. The area is sparsely populated and Outinen serves as a perfect example of a rural crossing in the Finnish context. The original setting of the crossing gates, as shown in the Figure 1. While the speed limit on the road was 80 km/h, this crossing was considered extremely dangerous as people did not slow down sufficiently to ensure they could stop before a train approached the crossing.

Page 122 GRADE-SEPERATION AT ROADRAIL CROSSINGS

Figure 14: Outinen crossing prior to changes.

The site has the following characteristics: Frequency of car traffic is low – 126 vehicles, with 95% private cars and 5% trucks. Number of trains – 13 passenger trains daily, an estimate of 7 freight trains, total of 20 daily trains There have not been any accidents at the crossing during the period of 1990-2000, so we estimate that despite the dangerous location of the crossing there is no annual monetarised safety impact of the grade separation. The average free speeds measured on the road were 61-66 km/h, the average crossing speeds were 44-48 km/h. Thus, the average slowdown at the crossing is 17-18 km/h. Providing the socio-economic profitability calculus for the crossing yields us the CBA results. A comparison of the net present values of the benefits (from both safety and mobility improvements) with the average cost of building a grade-separation, provides the benefit-cost ratio of as follows: 0.65 from using the approximate formula.

Page 123 GRADE-SEPERATION AT ROADRAIL CROSSINGS

5.3 CBA of an urban crossing

Crossing at Hennala is an urban rail-road crossing, which is situated on the railroad between Riihimäki and Kouvola at Lahti, a city with population of around 100,000 inhabitantis. The site has the following characteristics: Daily vehicle traffic – 4,328 vehicles, with 94% of private cars, 5% of trucks and 1% of buses; Number of trains per day – estimated at 70, with 53 passenger trains and 17 of freight trains. There average cost of the accident on the crossing is evaluated at 84094 euros, based on the fact that there were no severe accidents at the crossing during the period 1990-2000- The annual loss due to accidents, or the economic value of safety benefits due to implementation of the measure, is therefore equal to 8.410 euro at 2000 prices. The average free speeds measured on the road were 48-53 km/h, the average crossing speeds were 25-31 km/h. Thus, the average slowdown at the crossing is 36-40 km/h. Providing the socio-economic profitability calculus for the crossing yields us the CBA results. A comparison of the net present values of the benefits (from both safety and mobility improvements) with the average cost of building a grade-separation, provides the benefit-cost ratio of as follows: 0.25 from using the approximate formula.

6 Assessment Results

For the rural case, the CBA results yield a non-profitable CBA ratio of 0.65. This is in particular due to the savings in both average waiting times (which are abolished) and the increase in speed in the absence of level crossing as the accidents data did not support major savings from accident costs. The cost-benefit ratios for a grade-separation of the urban crossing was 0.25, which is not generally considered a profitable level for a project. However, given that there were no observed safety impacts to be added (which could be obtained from larger data of similar types of crossings to estimate the probability of severe accident and the associated monetary value) adding these elements to the case would most likely yield a higher benefit-cost ratio. The safety aspects play no role in economic analysis, due to the fact that there was only one minor accident at the urban crossing and none in the rural during the period 1990- 2000. However, one should remember that safety problems of the at-grade crossings are usually the main reason for consideration of grade-separation.

7 Decision Making Process

The study has utilised data from Finnish Rail Administration, which has an on-going evaluation program of railroad system, including level crossings. It is hoped that the results from economic analysis of safety measures could be applied in the future decision-making. For these purposes, the more analytical Israel model could be applied and, if needed, calibrated to fit the Finnish situation.

Page 124 GRADE-SEPERATION AT ROADRAIL CROSSINGS

The project's results – a list of crossings warranting a grade-separation, were adopted by the Planning Department of the Ministry of Transport, which is responsible for financing and planning of improvements of public road networks. The CBA provided a firm basis for the evalution's performance and for selecting crossings which have higher priorities for future investments.

8 Role of barriers

Considering the main groups of barriers to the use of EAT or to the implementation of evaluation results (WP2, 2004), one can conclude that none of them played a serious role in the project's performance. Authorities were very helpful in providing data and given that the analysis are ex post, in the sense that the projects were implemented without a CBA, the results are considered useful for future evaluations.

9 Discussion

A grade-separation of an at-grade crossing can be beneficial under certain conditions. The daily number of trains and daily road traffic volume are the main crossing parameters in this consideration as they influence both the accident frequencies and the extent of traffic delays, at the crossing. In some cases, conditions caused by weather or visibility consideration can create need to construct the level-crossing, even if the appears to be economically disadvantageous. In the study, the evaluation tools of standard road transport CBA were applied to grade- separation. Examples of a CBA of two typical crossings were provided. The crossings warrant a grade-separation while both safety and mobility benefits are accounted for in the rural setting, in the urban setting more detailed review of statistically meaningful safety impacts should be considered. In the rural context, the speed and delay impact starts to dominate the calculation when there are sufficient speed gains from the construction of the grade-separation. This is something that should be made clear as it may imply careless driving in the first instance. The CBA presented in this study was satisfactory from many viewpoints, such as: 1. the evaluation findings supported the measure's implementation, at least partially; 2. the evaluation performed was in line with the criteria of correct evaluation (WP3, 2004), as special data were collected for different evaluation tasks and statistical models were fitted to the data; 3. the accident costs were fitted to the accident type considered; 4. the evaluation study was initiated by the authorities and the results were accepted by the decision-makers. In general, in the case presented, the majority of technical and institutional barriers for the CBA's performance were overcome. It should be noted, though, that in general the railway crossings in Finland do not require black spot management, since the phenomena does not exist in Finland. Distribution of accidents is random and cannot be assigned to certain spots in the network. The evaluation results had a number of limitations, such as:

Page 125 GRADE-SEPERATION AT ROADRAIL CROSSINGS

1. The implementation costs include mostly initial investments. Maintenance costs were not explicitly considered neither for at-grade nor for grade-separated crossings. 2. The average value of implementation costs was applied for all sites considered. Providing specific values will need for detailed feasibility studies of specific locations. 3. No confidence interval was provided for the safety effect value. As explained previously, the safety effect in this case is stable (i.e. eliminating all accidents), whereas the safety benefits from the measure depend on the number of accidents expected at the site, per year. The latter was predicted by a model. However, safety impacts played almost no role in the analysis. 4. The contribution of safety factor to the benefits from the measure implementation was relatively low. This is the usual problem in calculating the socio-economic profitability of investment projects, where time savings dominate other impacts, including safety. These case studies suggest that the implementation of crossings is not related to safety assessment but other decision-making criteria. 5. Environmental impact was not quantified by the CBA performed.

Page 126 GRADE-SEPERATION AT ROADRAIL CROSSINGS

REFERENCES

Ahonen, T., A. Seise and E. Ritari (2003). Tasoristeysten turvallisuus Porin ympäristön rataosilla. Research Report RTE3815/03. 48 pages. VTT, Espoo. Ahonen, T., A. Seise and E. Ritari (2004). Tasoristeysten turvallisuus Seinäjoki-Kaskinen- rataosalla. Research Report RTE2208/04. 87 pages. VTT, Espoo. Ahonen, T., A. Seise and E. Ritari (2004). Tasoristeysten turvallisuus Seinäjoki-Oulu- rataosuudella. Research Report RTE742/04. 71 pages. VTT, Espoo. Ahonen, T., A. Seise and E. Ritari (2004). Tasoristeysten turvallisuus Pieksamäki- Joensuu-rataosuudella. Research Report RTE154/04. 68 pages. VTT, Espoo. Gitelman V., Hakkert A.S. (2001) Updating procedures for the consideration of grade- separation at road-rail crossings in Israel. Research Report No 285/2001, Transportation Research Institute, Haifa, Israel (in Hebrew). Hytönen, J., T. Ahonen and A. Seise (2004). Tasoristeysten turvallisuus Joensuu- Uimaharju-rataosuudella. . Research Report RTE2207/04. 47 pages. VTT, Espoo. Hytönen, J., T. Ahonen and A. Seise (2004). Tasoristeysten turvallisuus Niirala-Säkäniemi rataosalla. . Research Report RTE776/04. 39 pages. VTT, Espoo. Ministry of Transport and Telecommunications (2003). Guidelines for project appraisal. Ratahallintokeskus (2004) Suomen rautatietilasto 2004. The Finnish Railway Statistics. WP3 (2004) Improvements in efficiency assessment tools. ROSEBUD. WP2 (2004) Barriers to the use of efficiency assessment tools in road safety policy. ROSEBUD.

Page 127 CASE F2: Grade-separation at ROAD-RAIL crossings

Technion - Israel Institute of Technology Transportation Research Institute

ROSEBUD WP4 - CASE F REPORT

GRADE-SEPARATION AT ROAD-RAIL CROSSINGS

BY VICTORIA GITELMAN AND SHALOM HAKKERT,

TRANSPORTATION RESEARCH INSTITUTE, TECHNION, ISRAEL GRADE-SEPERATION AT ROADRAIL CROSSINGS

TABLE OF CONTENTS

1 PROBLEM ...... 131 2 DESCRIPTION OF MEASURE...... 131 3 TARGET GROUP ...... 132 4 ASSESSMENT METHOD...... 132 4.1 Assessment tools developed...... 132 4.1.1 Evaluating economic losses due to safety problems...... 132 4.1.2 Evaluating economic losses due to vehicle delays...... 134 4.2 Considering the cost of the measure...... 135 5 ASSESSMENT QUANTIFICATION...... 136 5.1 General ...... 136 5.2 CBA of a rural crossing ...... 137 5.3 CBA of an urban crossing ...... 138 6 ASSESSMENT RESULTS...... 138 7 DECISION-MAKING PROCESS...... 139 8 ROLE OF BARRIERS ...... 139 9 DISCUSSION...... 139

Page 129 GRADE-SEPERATION AT ROADRAIL CROSSINGS

CASE OVERVIEW

Measure Grade-separation of at-grade road-rail crossings Problem Train-vehicle collisions at the crossing (and vehicle delays due to crossing's closures) Target Group Train-vehicle accidents Targets Diminishing accidents and traffic delays Initiator Planning Department of the Ministry of Transport Decision-makers Planning Department of the Ministry of Transport, Road Authorities, Railway Authorities Costs Investments in grade-separation construction; paid by the Ministry of Transport Benefits The benefits are accident savings and a reduction in traffic delays. Driving public will benefit. Cost-Benefit Ratio For a rural crossing: from 1:1.9 to 1:2.8; for an urban crossing: from 1:1.0 to 1:1.4.

Page 130 GRADE-SEPERATION AT ROADRAIL CROSSINGS

1 Problem

The large majority of road-rail crossings in Israel, like in any country, are level (at-grade) crossings. In general, at-grade road-rail crossings are associated with economic losses due to vehicle delays and train-vehicle collisions. The problem becomes urgent when a rapid increase in train traffic occurs as happened with the Israeli railways since the mid 90s. This necessitated a policy concerning the need and priorities for grade separation at such crossings. To illustrate the situation in numbers over the past years (1995-2000), a rapid increase in train frequencies occurred on most railway lines in Israel: the total number of passenger trains per day changed from 80 to more than 200, with an annual average increase of 21 percent. This was accompanied by a jump in rail-highway crossing accidents in 1997- 1999. Besides, a steady increase in road traffic over the years took place, as well as the doubling of railway lines along the main train corridors. All these developments stimulated the Ministry of Transport to re-examine the state of road-rail crossing safety. A preliminary analysis demonstrated that a significant part of locations with accident occurrences are at-grade crossings that are equipped with automatic safety gates (i.e. have the highest form of safety protection for at-grade crossings). Due to high train frequencies and significant road traffic volumes at some crossings, the economic losses because of vehicle delays might be high. Therefore, the question was to point out the sites where a grade-separation is warranted. The process of grade-separation is expensive. It is therefore essential to provide a systematic approach for decision-makers that will lead to a considered decision on the benefits and costs associated with grade-separation. However, a detailed investigation of a specific crossing is time-consuming and costly (Tustin et al. 1986; Taggart et al. 1987), and cannot be reasonably performed for a large number of sites. Thus, at the initial stage, screening tools are required that will assist in choosing from the whole set of locations (i.e. from the whole railway network) those warranting further consideration. Therefore, the Ministry of Transport initiated a study to develop such screening tools and provide for an exhaustive list of sites having a potential for grade separation throughout the whole railway network The screening tools for crossings’ consideration for grade-separation were developed by the study Gitelman, Hakkert (2001). The tools consist of a safety model, a formula for estimating the economic loss due to vehicle delays at a crossing, and a qualification criterion. The tools are based on economic principles, comparing the economic loss due to an at-grade crossing with the average cost of grade-separation. In this report we will briefly discuss the development of the screening tools and present a cost-benefit analysis (CBA) of grade-separation of two representative road-rail crossings.

2 Description of measure

A grade-separation of a road-rail crossing means building a bridge or a tunnel instead of an existing at-grade crossing. A grade-separation eliminates existing railway-road crossings and consequently, removes the problem of train-vehicle collisions at the site considered. Besides, the grade-separation considerably diminishes the amount of road traffic delays at the site that previously stemmed from the crossing’s closures due to trains’ movements.

Page 131 GRADE-SEPERATION AT ROADRAIL CROSSINGS

A grade-separation is usually considered for road-rail crossings that are already protected by automatic gates.

3 Target Group

The target accident group are train-vehicle collisions at level crossings. The project aimed at developing screening tools for selecting crossings with a high potential for grade-separation, i.e. those crossings where the costs of vehicle delays and safety problems associated with the at-grade crossings are sufficiently high in order to justify building a grade-separation. Such tools are needed for decision-makers, as they both stimulate an objective policy and a systematic approach to the issue, and define a priority for grade separation at crossings. The tools are applied to perform a CBA of a grade-separation of two typical crossings.

4 Assessment method

4.1 Assessment tools developed

• The economic losses associated with the current situation, i.e. at-grade crossing, stem from two main factors: vehicle delays and safety problems. These losses represent the economic benefits, which can be attained due to eliminating the at-grade crossing. The CBA should compare these potential benefits with the costs of building a grade- separation. The assessment tools developed for estimating potential benefits from a grade-separation include (Gitelman, Hakkert, 2001): • An accident prediction model, which, along with accident costs, supplies a basis for evaluating losses due to safety problems at level crossings. • A model for evaluating economic losses due to vehicle delays at any crossing, based on its parameters. • A quantitative criterion for grade separation, which combines the results of both models. • Field measurements at twenty representative sites (out of more than 200), as well as accident data and the crossings’ inventory for five years (1995-1999), provided a basis for building the tools.

4.1.1 Evaluating economic losses due to safety problems

Safety concerns at level crossings frequently provide the main reason for grade-separation (Europe@ 1998; US GAO 1995). The evaluation of the safety factor needs two inputs: an estimate of the expected number of accidents per crossing and the cost of an average crossing accident. The expected number of accidents per crossing represents the annual

Page 132 GRADE-SEPERATION AT ROADRAIL CROSSINGS number of accidents that will be prevented due to grade-separation, whereas their costs demonstrate the economic value of safety benefits expected. Both inputs were developed based on the data on train-vehicle accidents, which occurred at all level Israeli crossings over the five years, 1995-1999. To estimate the number of accidents expected at a specific crossing based on the crossing characteristics, a multiple regression model was developed. The database for the models’ development comprised, in total, 80 accidents and 994 “crossing-years”; both populations were built as a unification of five-year statistics, with necessary updates of crossings’ characteristics for each year considered. The model was developed by means of the S+ and SAS statistical packages. The Poisson rather than the Negative Binomial distribution was found to fit the accident frequencies at the crossings. As known, when a Negative Binomial distribution is found to be most suitable, it is customary to apply the Empirical-Bayes method for predicting accident numbers (WP3, 2004). In our case, the expected number of accidents at a local crossing should be defined mainly by its type (i.e. estimated by means of the fitted regression model) without a need for further correction of the value using the Empirical-Bayes method. In other words, the expected number of accidents at a crossing is equal to the expected number of accidents at an average site of this type. The model recommended for application in Israeli conditions looks as follows: λ = exp(-5.904 + 1.183*PROTECT + 0.426*NVOL + 0.876*NTRAIN - 0.6*NTRAIN*PROTECT) (1) where λ = the expected number of accidents at a local crossing, per year; PROTECT= protection level, with 1 for “gate” or “lights”, 0 for “signs only”; NVOL = category of traffic volume, a number between 1-5 (see values in Table 50); NTRAIN = category of the number of trains, a number between 1-7 (see values in Table 50).

Table 50: Categories of crossing characteristics, for evaluating crossing safety Vehicle traffic volume Daily number of trains Category Value, thousand Category Value, trains per day number vehicles per day number 1 ≤ 1.0 1 Irregular* 2 1.0-5.0 2 ≤ 10 3 5.0-10.0 3 10-30 4 10.0-20.0 4 30-50 5 ≥ 20.0 5 50-80 6 80-110 7 ≥ 110 *does not appear in operative timetable The cost of an average crossing accident was estimated using actual accident consequences over the 5-year period. The list of accident consequences included the effects of human injury; damage to vehicle, train and crossing equipment; delays of road and train traffic; and the activities of authorities involved, i.e. police, trial, railway accident investigation team, etc – see Table 51. The cost of an average train-vehicle crossing accident was estimated to be about NIS 448,000 or € 118,000 (at 2000 prices). The accident injury and fatality costs were calculated on the basis of the gross loss of output

Page 133 GRADE-SEPERATION AT ROADRAIL CROSSINGS method. Had the ‘willingness-to-pay’ method been used, the accident costs would probably have doubled. Table 51: Calculation of Cost of Average Crossing Accident Ordinal Accident Consequence Frequency Unit Cost, Contribution to number per Accident NIS Total Cost, NIS 1. Fatality 0.091 2,726,500 248,112 2. Injury 0.334 205,000 68,470 3. Damage to vehicle 0.979 47,487 46,490 4. Damage to train 0.448 65,758 29,460 5. Damage to crossing equipment 0.113 27,782 3,139 6. Passenger train traffic delays 0.468 3,123 1,462 7. Freight train traffic delays 0.379 491 186 8. Railway maintenance work delay 0.091 240 22 9. Road traffic delays 1 5,861 5,861 10. Activities of authorities involved: 1 44,447 44,447 police; trial; social insurance, etc and railway accident investigation team Total accident cost NIS 447,649 or € 118,020 Note: NIS = New Israeli Shekel, € 1= 3.793 NIS. At 2000 prices.

The composition of the accident cost with the expected accident number supplies the annual cost evaluation of safety problems at a crossing.

4.1.2 Evaluating economic losses due to vehicle delays

• A sample of 20 crossings was selected for detailed measurement. The crossings were selected from among the busier lines of the rail network. • At each crossing, the parameters measured were as follows: vehicular traffic volumes, closure times and queue release times, and vehicle speeds. For traffic volumes, hourly distribution was attained, along with the traffic subdivision into three vehicle classes: cars, trucks and buses. Prior to the evaluation, the hours were divided into three time intervals, according to the traffic volumes observed: peak, low (night) and intermediate volumes. The closure times (of the automatic gates) were estimated for three types of trains: passenger trains, freight trains and operational trains. Vehicle speeds were measured at a “free” distance from the crossing and immediately before the crossing. Besides, for each crossing the number of train transitions per each hour was calculated based on the railway line operative timetable. • The analysis revealed that the closure times depended on the train type, train speed and the vicinity of station, whereas for the times for queue release no clear dependence was seen between this parameter and the average traffic volume or closure time [GITELMAN, HAKKERT, 2001].

Page 134 GRADE-SEPERATION AT ROADRAIL CROSSINGS

The annual cost of vehicular delays at a crossing was estimated from:

D = 260⋅[N ⋅ d1 + (V − N) ⋅ d2 ] (2) where D = annual cost of vehicular delays, NIS (for 260 working days a year), V = vehicular daily traffic volume, vehicles, N = number of vehicles stopped at the crossing per day, d1 = average cost of a vehicle’s stopping at the crossing, NIS, d2 = average cost of a vehicle’ slowing down at the crossing, NIS. The economic losses sustained from traffic delays ensue from additional consumption of fuel and other vehicle expenses and from the time lost to vehicle occupants because of “velocity cycles” when passing the crossing. Estimating d1 and d2 in the above formula, the losses due to different vehicle and train types at a specific crossing were weighted, in accordance with the shares of these types in daily vehicle/ train traffic at this site. The detailed calculation of vehicle delay costs at a specific crossing consists of various and multiple data considerations. Hence, for a rapid screening of sites, an approximate formula was developed which allows estimation based on the crossing’s parameters and without prolonged calculations. The model fitting was performed by means of the SAS multiple linear regression module where the parameters and estimates of the sample crossings served as a database. The approximate formula recommended for application was [GITELMAN, HAKKERT, 2001]: Y = -0.656044 + 0.000108*V + 0.0023038*Trains + 0.094042*Slowdown (3) where Y = annual economic loss due to vehicle delays at a crossing, million NIS, V = daily traffic volume (vehicles), Trains = daily number of trains (trains), Slowdown = average vehicle speed reduction due to a crossing (km/h).

4.2 Considering the cost of the measure

Summing up the costs of vehicle delays and of safety problems provides a value of annual economic loss at the level crossing, i.e. the magnitude of economic benefits, which could be attained due to a grade-separation. This value should be compared with the construction costs. Consultation with local economic experts and authorities that supervised some recent grade separations suggested that a figure of NIS 10 million (€ 2.6 million), at 2000 prices, would be a reasonable average for the construction of a grade separation at an Israeli crossing. Considering the net present value of the construction costs, with a 7 percent discount rate and a 15-year project life19 supplies a value of benefits (or annual economic loss at a level crossing) that would justify a grade-separation. This is a loss of 1.1 million NIS at least (€ 0.290 million; at 2000 prices), to provide a benefit-cost ratio higher than 1.

19 According to Israeli recommendations for economic evaluation of transport projects

Page 135 GRADE-SEPERATION AT ROADRAIL CROSSINGS

Applying the boundary value of 1.1 million NIS to the estimates of sample crossings, 13 sites (out of 20) were chosen as meriting a grade-separation [GITELMAN, HAKKERT, 2001]. Considering the parameters of two groups of crossings, i.e. those warranting and those not yet warranting a grade-separation, a criterion was developed for a preliminary crossing's qualification from the viewpoint of its potential for grade-separation. The criterion examines two crossing's parameters: daily vehicle traffic and number of trains per day, and compares them with the boundary values for urban or rural crossings (depending on the crossing's location). For example, for urban crossings, the consideration for a grade-separation is irrelevant for sites with less than 20 trains per day or when the daily vehicle traffic is less than 8,000 [GITELMAN, HAKKERT, 2001].

5 Assessment Quantification

5.1 General

In this section we consider a CBA of grade-separation of two typical crossings. To note, a cost-benefit and not cost-effectiveness analysis was chosen, due to following reasons: 1. multiple policy objectives to be considered (both safety and mobility), 2. monetary valuations of all benefits and costs should be applied (inter alia, to justify the implementation of the measure). The main data elements to be provided for the CBA performance are (WP3, 2004): • A definition of unit of implementation for the measure; • An estimate of the number of accidents expected to be prevented per unit implemented of the measure, through: identification of target accidents, estimate of the number of target accidents expected to occur per year, estimate of the safety effect of the measure on target accidents; • Accident costs; • Other monetary values depending on the effects considered; • An estimate of the costs of implementing the measure; • The economic frame for the evaluation (length of service life, interest rate). In our case of grade separation of at-grade crossings, the above data elements will be as follows: • The unit of implementation is one at-grade crossing; • Target accidents are all train-vehicle accidents at the at-grade crossings. The number of target accidents expected to occur per year can be estimated using a prediction model (formula 1 above). The safety effect of the measure is 100% reduction in target accidents, as a grade-separation implies the elimination of all train-vehicle collisions. Thus, in this case, the number of accidents expected to be prevented following implementation of the treatment is equal to the number of target accidents that are expected to occur at the site, prior to implementation of the treatment.

Page 136 GRADE-SEPERATION AT ROADRAIL CROSSINGS

• Accident costs – see Section 4.1.1; • Other monetary values include costs of travel time and vehicle operating costs; they can be estimated using formulae from Section 4.1.2; • The average cost of implementing the measure – see Section 4.2; • The economic frame for the evaluation: 15-year project life, with 7% discount rate. The accumulated discount factor is 9.108. The CBA is performed for two at -grade crossings: No 19 and No 133.

5.2 CBA of a rural crossing

Crossing No. 19 is a rural road-rail crossing, which is situated on 45.106 km of Haifa-Tel- Aviv railway line and on a regional road No. 651; the crossing is protected by automatic gates. The site has the following characteristics: Daily vehicle traffic - 15,330 vehicles, with 93% of private cars, 5% of trucks and 2% of buses; Number of trains per day - 132, with 89% of passenger trains and 11% of freight trains. Using Formula 1, the expected number of accidents per year will be 0.338 (that is the number of accidents to be prevented due to the measure). The annual loss due to accidents, or the economic value of safety benefits due to implementation of the measure, is equal to 0.151 million NIS (€ 0.040 million), at 2000 prices. The average free speeds measured on the road were 66-68 kph, the average crossing speeds – 51-52 kph (for private cars, buses) and 44 kph (for trucks). Thus, the average slowdown at the crossing is 15-16 kph (for private cars, buses) and 22 kph (for trucks). The average cost of a slowdown at the crossing is estimated to be 0.49 NIS. The average length of the crossing closure is 0.37 min due to a passenger train, and 1.46 min due to a freight train. The average cost of stopping due to the crossing's closure is 2.23 NIS. Using Formula 2 for a detailed calculation, the annual costs of vehicle delays at the crossing will be 2.916 million NIS (€ 0.769 million), at 2000 prices. Another estimate of the annual costs of vehicle delays, based on the approximate Formula 3 will be 1.925 million NIS (€ 0.508 million), at 2000 prices. Effects, safety, and mobility compose the benefits from the grade-separation of the crossing. A comparison of the net present values of the benefits with the average cost of building a grade-separation provides the cost-benefit ratios as follows: 1:2.79 when the costs of delays come from a detailed calculation; 1:1.89 when the costs of delays come from the approximate formula.

Page 137 GRADE-SEPERATION AT ROADRAIL CROSSINGS

5.3 CBA of an urban crossing

Crossing No. 133 is an urban railroad crossing, which is situated on 114.806 km of Remez Junction-Kiriam Gat railway line and on Jabotinsky Street in Beer Yakov; the crossing is protected by automatic gates. The site has the following characteristics: Daily vehicle traffic - 13,156 vehicles, with 91% of private cars, 7% of trucks and 2% of buses; Number of trains per day - 71, with 86% of passenger trains and 14% of freight trains. Using Formula 1, the expected number of accidents per year will be 0.195 (that is the number of accidents to be prevented due to the measure). The annual loss due to accidents, or the economic value of safety benefits due to implementation of the measure, is equal to 0.087 million NIS (€ 0.023 million), at 2000 prices. The average free speeds measured on the road were 43-49 kph, the average crossing speeds – 42-47 kph. Thus, the average slowdown at the crossing is 1-2 kph. The average cost of a slowdown at the crossing is 0.13 NIS. The average length of the crossing closure is 0.51 min due to a passenger train, and 0.77 min due to a freight train. The average cost of a vehicle’s stopping due to the crossing's closure is 1.78 NIS. Using Formula 2 for a detailed calculation, the annual costs of vehicle delays at the crossing will be 1.023 million NIS (€ 0.270 million), at 2000 prices. Another estimate of the annual costs of vehicle delays, based on the approximate Formula 3, will be 1.490 million NIS (€ 0.393 million), at 2000 prices. A comparison of the net present values of the benefits (from both safety and mobility improvements) with the average cost of building a grade-separation, provides the cost- benefit ratios as follows: 1:1.01 when the costs of delays come from a detailed calculation; 1:1.44 when the costs of delays come from the approximate formula.

6 Assessment Results

The cost-benefit ratio for a grade-separation of crossing No. 19 ranges from 1:1.9 to 1:2.8; the cost-benefit ratio for a grade-separation of crossing No. 133 – from 1:1.0 to 1:1.4. In both cases, the treatment is warranted from the economic viewpoint. The safety factor had only a minor contribution to the economic benefits expected: 4.9%- 7.3% for crossing No. 19, 5.5%-7.8% for crossing No. 133. However, one should remember that safety problems of the at-grade crossings are usually the main reason for consideration of grade-separation. Applying the evaluation tools developed for the examination of all existing Israeli crossings, in the year 2000, 30 sites out of 216 were found to warrant a grade-separation (Gitelman, Hakkert, 2001).

Page 138 GRADE-SEPERATION AT ROADRAIL CROSSINGS

7 Decision-Making Process

The study was initiated by the Planning Department of the Ministry of Transport in co- operation with the Israeli Railways. The study's steering committee included decision- makers having senior positions in the Ministry of Transport and in the Railway Authority. The project's results – a list of crossings warranting a grade-separation were adopted by the Planning Department of the Ministry of Transport, which is responsible for financing and planning of improvements of public road networks. The CBA provided a firm basis for the evaluation’s performance and for selecting crossings that have higher priorities for future investments.

8 Role of barriers

• Considering the main groups of barriers to the use of EAT or to the implementation of evaluation results (WP2, 2004), one can conclude that none of them played a serious role in the project's performance. The transport authorities initiated the project and promoted the implementation of its results, therefore indicating that institutional or implementation barriers are actually irrelevant in this case. • The technical barriers, e.g. lack of knowledge of safety effect or of accident costs, existed at the beginning, but were solved later by means of relevant data collection and fitting statistical models for various evaluation needs. The assistance by railway executives was extremely important at the stage of collecting data on train-vehicle accidents and on railroad crossings' characteristics.

9 Discussion

A grade-separation of an at-grade crossing can be beneficial under certain conditions. The daily number of trains and daily road traffic volume are the main crossing parameters in this consideration as they influence both the accident frequencies and the extent of traffic delays at the crossing. In the study, the evaluation tools for preliminary CBA of a grade-separation were developed and applied for selecting crossings warranting implementation of the measure. Examples of a CBA of two typical crossings are provided. The crossings warrant a grade- separation, while both safety and mobility benefits are accounted for. The CBA presented in this study was satisfactory from many viewpoints, such as: 1. the evaluation findings supported the measure's implementation; 2. the evaluation performed was in line with the criteria of correct evaluation (WP3, 2004); in particular, special data were collected for different evaluation tasks and statistical models were fitted to the data; 3. the accident costs were fitted to the accident type considered; 4. the evaluation study was initiated by the authorities and the results were accepted by the decision-makers.

Page 139 GRADE-SEPERATION AT ROADRAIL CROSSINGS

In general, in the case presented, the majority of technical and institutional barriers for the CBA's performance were overcome. The evaluation results had a number of limitations, such as: 1. The implementation costs include mostly initial investments. Maintenance costs were not explicitly considered either for at-grade or for grade-separated crossings. 2. The average value of implementation costs was applied for all sites considered. Providing specific values will need detailed feasibility studies of specific locations. 3. No confidence interval was provided for the safety effect value. As explained previously, the safety effect in this case is stable (i.e. eliminating all accidents), whereas the safety benefits from the measure depend on the number of accidents expected at the site per year. The latter was predicted by a model. 4. The contribution of a safety factor to the benefits from the measure implementation was relatively low. This contribution might be doubled had the ‘willingness-to-pay’ method been used for estimating accident costs. 5. Environmental impact was not quantified by the CBA performed.

References

Europe’s Approach to Rail Crossing Safety (1998): ITE Journal, Feb.,18. GITELMAN V., HAKKERT A.S. (2001): Updating procedures for the consideration of grade-separation at road-rail crossings in Israel. Research Report No 285/2001, Transportation Research Institute, Haifa, Israel (in Hebrew). Taggart, R.C., LAURIA, P. et al. (1987): Evaluating Grade-Separated Rail and Highway Crossing Alternatives. NCHRP Report 288, Transportation Research Board, Washington D.C. TUSTIN, B.H., RICHARDS, H., MCGEE, H. and PATTERSON, R. (1986): Railroad- Highway Grade Crossing Handbook. Report No. FHWA TS-86-215, Springfield VA. United States General Accounting Office (US GAO) (1995): Status of Efforts to Improve Railroad Crossing Safety. Report GAO-RCED-95-191, Washington, D.C. WP3 (2004): Improvements in efficiency assessment tools. ROSEBUD. WP2 (2004): Barriers to the use of efficiency assessment tools in road safety policy. ROSEBUD.

Page 140 CASE G: MEASURE against collisions with trees

ROSEBUD WP4 - CASE G REPORT

MEASURES AGAINST COLLISIONS WITH TREES RN134 (LANDES) FRANCE

BY PHILIPPE LEJEUNE,

CETE SO, FRANCE MEASURE AGAINST COLLISIONS WITH TREES

TABLE OF CONTENTS

1 CASE OVERVIEW...... 143 2 PROBLEM TO SOLVE ...... 145 3 DESCRIPTION OF THE MEASURE...... 146 4 TARGET ACCIDENT GROUP...... 146 5 ASSESSMENT METHOD...... 147 5.1 Choice of CBA...... 147 5.2 Assessment tool...... 147 5.3 Road safety of the collisions against trees ...... 149 5.4 Type of assessed impacts...... 149 5.5 Costs of the measure ...... 150 5.6 Costs of accidents...... 150 6 ASSESSMENT QUANTIFICATION...... 151 7 ASSESSMENT RESULTS...... 152 8 DECISION MAKING PROCESS...... 152 9 IMPLEMENTATION BARRIERS ...... 153 10 CONCLUSION ...... 154

Page 142 MEASURE AGAINST COLLISIONS WITH TREES

Case Overview

Measure The measure aims to avoid the collisions with the trees along 26.5 km of the national road RN 134 over the "Département des Landes" in the Southwest of France. The measure consists of the implementation of 7800 meters of guardrails, 13 frontage accesses and 8 lay-by. Problem Some stretches of the road RN 134 crossing through the forest have a high level of risk in terms of crashes and severity due to the row of trees along the road side. The problem was to propose and negotiate measures to reduce the number and the severity of the crashes by ensuring the protection of the row of trees by the means of guardrails when it was possible, or otherwise by means of tree felling. Target Group All the road users driving on two stretches of the national road RN 134, which had a high level of risk of collision with trees. Targets The safety measures applied along the tree-lined stretches of road had two main objectives: 1) Avoid the collisions of the vehicles against the trees, and 2) Reduce the accident severity of the remaining crashes. This second objective implies the use of normalized guardrails insuring the vehicles against violent impact, throwing the vehicles to the opposite carriageway. Initiator The initiator of this local safety road improvement is the local transport administration (DDE-CDES), but other actors at the French national, regional and local levels are also involved in decision-making and funding. Decision-makers Mid-level civil servants of the local transport administration (DDE-CDES) make the choices among the panel and define the time schedule of the "accepted" local road safety measures. These measures are mentioned in a ministerial decision signed by high-level civil servants of the Road Directorate of the Ministry of Transport at the national, regional, and local levels.

Page 143 MEASURE AGAINST COLLISIONS WITH TREES

Costs The total cost for implementing the measure was around 1 million €, including management, studies, implementation and site supervision. All these costs have been paid by the Ministry of Transport through the financial management of the regional administration. Benefits The main benefit from implementing the measure consists of an important reduction of the number of accidents against trees, fatalities and crash severity. Cost-Benefit Ratio The Cost/Benefit ratio is 8.69.

Page 144 MEASURE AGAINST COLLISIONS WITH TREES

1 Problem

The RN 134, which crosses the forest of “Landes” along 64.5 km, has long, tree-lined stretches of road on which before the measure, 38.5% of the accidents occurred against trees. A detailed traffic safety study showed that 58% of the accidents occurred over two stretches of road, which is 26.5 km length. Finally, the survey shows that 82% of the accidents against trees on the RN 134 occurred alongside the 26.5 km of these two stretches of road. Furthermore, during the period before the treatment (1993-1997) the safety indicators (accidents, casualties and injuries) of the crashes against trees were increasing (see Figure 15). Figure 15: Indicators and trends of accidents against trees

Indicators and trends of accidents against trees (stretches of road RN134 before treatments) 10

8

6

4

2

0 1993 1994 1995 1996 1997 Accidents Killed Injured Seriously

On the other hand during the same period (1993-1997) the safety indicators (accidents, casualties and injuries) of crashes against trees were decreasing alongside the roads of Landes (see Figure 16 below). Figure 16: Safety of crashes against trees in Landes

LANDES Safety of crashes against trees 120

100

80

60

40

20

0 1993 1994 1995 1996 1997 Accidents Killed Injured Seriously

Therefore the problem was to take measures to reduce the number and the severity of the crashes alongside these 26.5 km, which had the highest and increasing level of risk. For this purpose, the more suitable measures were the protection of tree rows by means of

Page 145 MEASURE AGAINST COLLISIONS WITH TREES guardrails, when possible, or otherwise tree felling should be examined. This second measure raised other difficulties due to ecological pressure groups that are against tree felling. Therefore, it has been decided to spend time and money to reach an agreement between the local authorities, decision-makers and ecological pressure groups to solve the problem jointly, i.e. traffic safety taking into account the ecological aspect.

2 Description of the measure

Taking into account this road safety problem specifics, i.e. traffic safety and ecology, the study carried out by the local administration of the Ministry of Transport (DDE-CDES) located precisely the stretches of road to be treated. Along all 26.5 km of these road stretches, the choice between guardrails and cutting trees down had to be done case by case according to the five following technical criteria: 1. The distance between the trees and the carriageway, 2. The number of trees (isolated trees to cut down), 3. Lay-by where hard shoulders are missing, 4. General state of health of trees, 5. The frontage accesses to be remained. In each case the decision was taken according to these criteria, keeping in mind the ecological aspects; therefore, the following measures have been performed: • 7800 meters of guardrails have been implemented where the preserved trees put the road users’ lives at risk, • 8 lay-by (emergency stop facilities) have been implemented at regular intervals where the hard shoulders were missing due to the narrow land and guardrails implementation, • 13 frontage accesses remained. The different steps to perform this measure were: • Management of the road safety measure and report related to the ecological topic, • Implementation plans: topographical surveying, choices between guardrails and tree felling, frontage access treatments project report etc., • Installation of the safety measure guardrail implementation, tree felling, road equipments and frontage access treatment, • Site supervision.

3 Target accident group

The road safety stakes in terms of accidents, casualties and injuries of the crashes against trees are summarised on the following figures. The target accident group involved those in crashes against trees and the related severity on the 26.5 km stretches of the RN 134.

Page 146 MEASURE AGAINST COLLISIONS WITH TREES

Figure 17: Treated sections of crashes against trees

Treated sections Safety evolution of crashes against trees 10

8

6

4 works

2

0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

Accidents Killed Injured Seriously Injured Slightly

The Figure 17 shows the impact of the measure on the safety indicators measured before and after the treatment of the row of trees along the roadside.

4 Assessment method

4.1 Choice of CBA

According to the theoretical principle of CBA as mentioned in the WP3 report, "CBA evaluates the economic benefits and costs of the objective….It aims to find if the proposed objective is economically efficient at all and how efficient it is". Taking into account the before-after data availability related to this traffic safety measure (accidents, traffic volumes, accident trends), CBA has been chosen for the assessment.

4.2 Assessment tool

The CBA ratio defined as: Present value of all benefits Benefit-cost ratio = Present value of implementation costs

All the following data have been collected during the periods before and after: 1. Accidents 2. Casualties 3. Injuries (severe and slight) according to current French definitions 4. Traffic volumes These data have been collected on the treated stretches of road and on reference areas before and after the measure implementation. The present value of all benefits has been

Page 147 MEASURE AGAINST COLLISIONS WITH TREES calculated from the safety impacts listed above (accidents causalities, and injuries), taking into account the trends of each of these variables in order to assess the numbers of accidents, casualties and injuries prevented. By this way it has been possible to apply the fundamental principle of the methodology proposed by Ezra HAUER20 "…to assess the effect of a treatment on the safety of some entity, one has to compare what would have been the safety of the entity in the after period had treatment not be applied, to what the safety of the treated entity in the after period was". This fundamental principle leads to assess "what would have been the safety (accidents, casualties and injuries) of the crashes against trees in the after period if the measure (guardrails or tree felling) had not been applied". In accordance with this principle, the impact of the measures have been calculated by comparing the counted "before safety values" to the assessed Ho safety values which are the "after period safety values" assessed under the Ho Hypothesis (i.e. if the measures had not been applied). The method 2 consists in calculating the theoretical "after accident numbers" as follows:

Theoretical After Accident Numbers = Po x (N after + N before) Where: Po is the probability of accidents under Ho i.e. if the measures had not been applied N after and N before is the number of accidents counted on the treated section after and before the implementation of the measure. Due to the random properties of the number of accidents, Po is calculated as follows: Evol × Daf × Taf Po = (Evol × Daf × Taf ) + Dbf × Tbf

where: • Evol: is the trend of the analysed traffic safety indicator calculated on a reference area (here the “Département des Landes”) • Daf and Dbf are the duration of the periods "after" and "before" the works concerning the implementation of the measure (assessment periods) • Taf and Tbf are traffic counted on the treated stretches of road "after" and "before" the works concerning the implementation of the measure (assessment periods)

20 Ezra HAUER "Observational Before-After Studies in Road Safety" Pergamon 1997 2 "Statistiques pour la Sécurité Routière" SETRA février 1999

Page 148 MEASURE AGAINST COLLISIONS WITH TREES

4.3 Road safety of the collisions against trees

The road safety indicators of the collisions against trees can be summarised as follows: Table 52: Safety results

Treated strechtes Landes of road Before After Before After 1993 1999 1993 1999 trends to to to to 1997 2003 1997 2003 Accidents 50 10 4 370 3 311 0,76

Killed 20 2 530 402 0,76 All crashes seriously 37 6 2 148 1 368 0,64 injured slightly 34 17 3 996 3 475 0,87 injured

Accidents 27 1 436 294 0,67

crashes Killed 11 0 129 85 0,66 against seriously trees 18 2 253 152 0,60 injured slightly 9 0 261 194 0,74 injured

4.4 Type of assessed impacts

As shown above in § 3, the measure had a significant impact on the traffic safety related to the crashes against trees. This impact was assessed in accordance with the "Before-After" assessment tool presented. Therefore according the above formulas, the prevented impacts have been calculated as follows:

(Prevented Accidents) = (Before Accidents) - Ho (After Accidents) (Prevented casualties) = (Before casualties) - Ho (After casualties) (Prevented injuries) = (Before injuries) - Ho (After injuries) The cost-benefit ratio has been calculated by using the accidents and casualties’ monetary values currently applied for the French CBA assessments. These road safety results related to the collision with trees have been applied according to the assessment tool described above and lead to the figures shown in the following tables.

Page 149 MEASURE AGAINST COLLISIONS WITH TREES

Table 53: Measure – safety impacts

Treated strechtes of road

Theoretical benefits Accident Number of accident of the probability P0 accidents prevented measure under Ho under Ho K€

Accidents 0,387 10,83 16,2 88,9

Killed 0,381 4,19 6,8 6 806,1 crashes against trees seriously 0,360 7,19 10,8 1 620,8 injured

slightly 0,410 3,69 5,3 116,8 injured

Total 8632,6

4.5 Costs of the measure

The costs for implementing the measure against collisions with trees were divided up in the following way: • Management of the road safety measure and report related to the ecological topic, • Implementation plans: topographical surveying, choices between guardrails and tree felling, frontage access treatments project report etc., • Installation of the safety measure guardrail implementation, tree felling, road equipments and frontage access treatment, • Site supervision. The total implementation costs were 993K €, which was paid by the Ministry of Transport through the financial management of the regional administration.

4.6 Costs of accidents

In France the cost of road safety has been assessed by Mr. Le NET (ENPC Paris) in a study 3carried out in 1991-1992 in which the different components of the price of human life have been calculated. This calculation applied the method called "Compensated Human Capital" using the following "marketed" and" non-marketed" costs. • Direct marketed costs,

3 "Prix de la vie humaine, application à l'évaluation du coût économique de l'insécurité routière" M. Le NET (ENPC) 1992

Page 150 MEASURE AGAINST COLLISIONS WITH TREES

• Medical and social costs, • Property damage costs (vehicles public equipments and environmental damages, fuel consumption, towing, etc., • Overheads as costs of police, justice, insurance services, etc., • Indirect marketed costs, • Costs of the loss of future productive capacity of fatalities and injuries, or jailed people, • Costs of the loss of future potential production, • Non-marketed costs; these costs are based on insurance company jurisprudence: o Cost of a killed person (moral wrong, prétuim mortis) o Cost of an injured person (prétium doloris) In 1999, this method led to the following costs: 1. Killed: 3950 KF (for which there are 88% of indirect marketed costs) 2. Seriously injured: 407 KF 3. Slightly injured: 86 KF 4. Property damages: 22 KF

These values have been updated in 2000 taking into account other country accident cost methodologies and including the correlation between the human life cost and GDP (Gross Domestic Product) per person. These updated costs (see the following §5) have been used for the present assessment.

5 Assessment Quantification

The quantitative analysis is a "case study" for which the data gathering and processing has been performed as follows: • This is a before/after study concerning the safety of the crashes with trees for which the data collected concerns all the accidents, casualties, injuries and the traffic volumes on the treated stretches of road and reference area, i.e. the road of the "Département des Landes". These figures have been cheeked and corrected, if necessary, at the local level according to police reports. The reference areas exclude the treated stretches of road. • Data sources are the official local accident statistic and traffic volumes that count ADT (Average Daily Traffic). The safety data concerns all the accidents involving at least one injured person, as defined below. Crashes against trees were identified. • Disaggregated data has been used; concerning the safety data, all details included in the accident database were available.

Page 151 MEASURE AGAINST COLLISIONS WITH TREES

• The time periods of the analysis are 1993 to 1997 for the “before period” and 1999 to 2003 for the “after period”. The period of construction (1998) has been cancelled from the data used. • Concerning safety data, killed and injured users have been defined according to current French definitions, i.e. six days for fatalities; hospitalised more than 6 days for seriously injured, and less than 6 days for the slightly injured. In each accident, the casualties, severe and slight injuries, and property damages were taking into account for the monetary valuations of the relevant measure impacts. • The source of monetary costs of accidents, casualties and injuries are those currently used in France4. For this assessment the costs for 2000 (safety and implementation) have been chosen as the reference year. This choice is due to the fact that the correlation between the safety costs and the GDP (Gross Domestic Product) has been used for the first time to make them comparable at the international level. These costs are: 5. Killed 1 000 K€ 6. Seriously injured 150 K€ 7. Slightly injured 22 K€ 8. Property damages 5.5 K€ The quantified safety results are summarised in the following table.

6 Assessment Results

Taking into account the safety parameters presented above, the "after" safety values have been calculated, and the impact of the measure has been assessed from the figures summarised in the Table 1 below. The total value of benefits is 8633 K€. The total value of implementation costs is 993 K€ Therefore according the above figures, the assessment tool (see § 4.2) and the French monetary valuation used, the cost-benefit ratio gives the following result: Present value of all benefits = 8633K€ / 993K€ . = 8.69 Present value of implementation costs

7 Decision-Making Process

The initiative of this local safety road improvement involved different actors at the French national, regional and local levels in terms of decision-making and funding.

4 Sécurité Routière en France" Bilan 2003 ONISR (Observatoire National Interministériel de Sécurité Routière) Documentation Française 2004

Page 152 MEASURE AGAINST COLLISIONS WITH TREES

The local safety measure described in this case report is part of a national French road safety program PRAS (Programme Régional d'Aménagements de Sécurité - Regional Road Safety program). This program is elaborated as follows: • Local road safety analyses are performed by the local transport administrations (DDE), which provide reports (Etude des Enjeux). These reports propose a set of breeding grounds of local road safety measures corresponding with the local road safety context. • All these reports are put together at the regional transport administration level (DRE), which performs a regional comprehensive road safety study. This comprehensive study is sent to the National Road Administration (Road Directorate), which in charge of the choices according to several criteria (political, financial, technical, etc). • The "accepted" local safety measures are mentioned in a ministerial decision signed by the three decision levels of the Transport Administration (national, regional and local DR, DRE, DDE), where applicable. • Funding is managed at the regional level (DRE). The local transport administration (DDE-CDES) makes the final choices among the "accepted" local road safety measures mentioned in the ministerial decision. The same local administration is in charge of the implementation work programs, time schedules, and so on, of these local road safety measures. These tasks have been undertaken, including dialogues with the local authorities and pressure groups, e.g. ecologists who are keeping a close watch on the measures leading to tree felling. For this purpose, an extra report has been written and provided to the local Commission of Sites (Commission des sites). This report presented a detailed study concerning the tree species of the ‘Landes’ forest and proposed compensating measures, which planned to replant trees in appropriate places. The Commission of Sites gave its approval. Otherwise, the implementation of this road safety measure would be impossible. Relevant decisions have been taken by the local transport administration and approved by the local authorities and pressure groups.

8 Implementation barriers

No significant technical barriers or difficulties have been met by the local decision-maker (DDE-CDES) in charge of the implementation of the measure. Only some problems related to the frontage access and the underground telecommunication cable networks were raised and were solved. The significant barriers before starting the measure were related to long and complicated administrative and financial procedures. These procedures involved several steps from the national (Road Directorate DR) to the regional (DRE) and local DDE/CDES) levels. Also, the local decision-maker in charge of the implementation of the measure has been waiting for the credit line before starting any work on the program.

Page 153 MEASURE AGAINST COLLISIONS WITH TREES

Concerning the assessment process, no barrier occurred. Local and national decision- makers and data providers provided all the necessary information and figures to perform the CBA.

9 Conclusion

The implementation of this road safety measure dealing with collision against the trees and the related “ex-post” cost-benefit analysis can be summarized as follows: • Although the problem to solve was included in a global and national traffic safety program it has been clearly identified and delimited. For this purpose local surveys have been integrated into the national road safety framework policy, decision processes, technical approaches and financing. The final decisions concerning the implementation of the measure are made by the local decision-maker from the administration (DDE-CDES). • In spite of the difficult technical choices and decisions to be made, in particular those related to political and environmental aspects linked to this measure, an agreement has been reached by means of dialogues involving the different local authorities, administrations, road engineers and pressure groups, e.g. ecologists. • Finally, the measure has been quite well accepted and the assessment shows good efficiency in terms of accidents and severity with a 8.69 cost-benefit ratio. • Therefore it seems that such an “ex-post” cost-benefit analysis could be an efficient in- put for further cost-benefit analyses (CEAs) and should be considered as only one of the decision-making process criteria to be used by the decision-makers to choose among available measures related to collisions against trees and side obstacles. The question is what will be the weight of such a CBA in the final decision-making process toward the other criteria to be taken into account by the decision-makers. This point should be discussed during the workshop and the conference.

References

(1) Ezra HAUER "Observational Before-After Studies in Road Safety", Pergamon, 1997 (2) Statistiques pour la Sécurité Routière" SETRA février 1999 (3) " M. Le NET (ENPC) "Prix de la vie humaine, application à l'évaluation du coût économique de l'insécurité routière" Ministère des Transports, 1992 (4) Sécurité Routière en France" Bilan 2003 ONISR (Observatoire National Interministériel de Sécurité Routière) Documentation Française, 2004

Page 154 CASE H: introducing signal control at a rural junction

Technion - Israel Institute of Technology Transportation Research Institute

ROSEBUD WP4 - CASE H REPORT

INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

BY VICTORIA GITELMAN AND SHALOM HAKKERT,

TRANSPORTATION RESEARCH INSTITUTE, TECHNION, ISRAEL INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

TABLE OF CONTENTS

1 PROBLEM ...... 158 2 DESCRIPTION OF MEASURE...... 159 2.1 General ...... 159 2.2 Current installation ...... 159 3 TARGET ACCIDENT GROUP...... 159 4 ASSESSMENT TOOLS ...... 160 4.1 Method for estimating safety effect ...... 160 4.2 Safety effect of introducing traffic signal control...... 162 4.3 Accident costs...... 163 5 COST-BENEFIT ANALYSIS...... 164 5.1 General ...... 164 5.2 Values of costs and benefits ...... 164 5.3 Cost-Benefit Ratio ...... 165 6 DECISION-MAKING PROCESS...... 165 7 DISCUSSION...... 166

Page 156 INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

CASE OVERVIEW

Measure Introducing traffic signal control at a rural junction Problem Traffic delays and accident occurrences due to conflict vehicle movements at a junction with no signal Target Group All injury accidents at the treated junction Targets Reducing traffic delays and the number of injury accidents at the junction Initiator Road authority – for the measure’s application; Ministry of Transport – for the evaluation of safety effect Decision-makers Road authorities, Ministry of Transport Costs Traffic lights’ design and installation, and the junction's realignment costs; paid by the Road Authority and the Ministry of Transport Benefits Estimated benefits stem from the expected savings in injury accidents at the treated junction. Benefits from reduced traffic delays are expected but not estimated. The driving public will benefit. Cost-Benefit Ratio 1:1.25, where the CBR accounts for safety effect only

Page 157 INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

1 Problem

In Israel, some 10% of both injury accidents and fatalities occur at rural junctions (CBS, 2003). When the accidents are observed at unsignalised intersections, the majority of accidents are usually right-angle, rear-end and pedestrian accidents. For unsignalised intersections, introducing traffic lights is frequently suggested as a safety treatment to reduce all accident types. International experience demonstrates (Elvik and Vaa, 2004) that the effect on accidents of traffic signal control at intersections was mostly positive, providing on average a 15% accident reduction at T-junctions and a 30% accident reduction at crossroads. At the same time, one should remember that the function of traffic lights is to provide time separation between conflicting traffic flows. Thus, the main purpose of introducing traffic lights at a junction is in improving traffic flows through the junction, i.e. in reducing delays, better use of the road’s capacity, providing successive traffic flows on arterial roads, etc. Eliminating conflicts between different traffic flows at the junction diminishes the probability of collisions and, therefore, may provide an additional benefit from traffic signal control – accident reduction. However, as it was proven by a number of studies when traffic lights are introduced at junctions with low traffic volumes, neither reductions in traffic delays nor safety benefits are usually observed. In some cases, deterioration in both conditions (i.e. an increase in traffic delays and accidents) was even reported. Therefore, the current Israeli guidelines on the design of traffic signal control recommend considering the introduction of traffic lights only for junctions with reasonably high traffic volumes (Ministry of Transport, 1981). The warrants for introducing traffic lights at a junction consider mostly the traffic volumes on the main and secondary roads, but enable also to account for additional conditions such as high accident numbers due to priority problems, lacking visibility, high approaching speed, or other geometric problems at the junction. The Israeli guidelines dictate a threshold of at least 10,000 private car units (or equivalent vehicle units) which enter the junction during the eight most heavily travelled hours, from both main and secondary roads, whereas the number of vehicles entering from the secondary road should be over 1,500. If the traffic volumes at a junction satisfy this demand, the introduction of traffic signal control can be considered. Presence of additional conditions (high accident frequencies, geometric problems, etc) may facilitate the above demand by up to 30% (Ministry of Transport, 1981). Prior to the installation of traffic lights the guidelines recommend considering other improvements such as priority signs, better visibility distances, road marking improvements, rumble bars to warn on approaching a junction, physical separation between different flows, pedestrian islands, etc. Such improvements are known as low- cost safety measures and are usually applied to sites with low to medium traffic volumes, but evident safety problems. Traffic lights’ installation is considered for sites with relatively high traffic volumes, which are close to the warrant’s demand. Possible safety benefits may be estimated in association with this infrastructure improvement, however, they will usually be treated as an additional benefit and never present the main reason for the application of the measure.

Page 158 INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

2 Description of measure

2.1 General

A road junction presents a natural point of potential conflict between different traffic streams. As traffic volumes increase, the probability of conflict increases too, and traffic delays worsen. Traffic signal control at intersection separates different traffic streams from each other and therefore improves the flow of traffic at the intersection and reduces accident occurrences. Traffic signal control is introduced using lights, which may be either time-controlled (phases change after a given time irrespective of the amount of traffic) or vehicle-actuated (the length of the phases is adapted to the amount of vehicles up to a given maximum phase length). The safety measure evaluated in the current study is the introduction of traffic signal control at a rural junction, which was previously controlled by priority signs, i.e. was an unsignalised intersection. The treatment is complex, including both the installation of traffic lights and the junction’s realignment. The latter typically includes arranging turning lanes, adding traffic islands, and improving signing and road marking at the site and in its vicinity.

2.2 Current installation

In the current study, we consider the installation of traffic signal control at a typical rural road junction, which is situated on a single-carriageway road. The junction is four-legged (a crossroad) with relatively high traffic volumes on the main road. The daily traffic volumes are: 9,000 vehicles entering the junction from the both directions of the main road and 2,000 vehicles – from the both directions of the secondary road. In total, nine injury accidents were observed at the junction over the three years prior to the traffic lights’ installation, whereby eight of them were associated with priority problems. The analysis of "before" traffic flows demonstrates that based on the traffic volumes only, the site would not satisfy the warrant for signal control’s installation. However, an additional consideration of accident records at the site enables to treat it as a boundary case warranting the measure. The purpose of the installation was, first of all, to improve the traffic flows and, possibly, to improve the site’s safety. The case is considered for the year 2002.

3 Target Accident Group

Considering the introduction of traffic signal control, the safety effect usually refers to all injury accidents (e.g. Elvik and Vaa, 2004). The positive effect is usually expected on right- angle accidents, other collisions from conflicting crossing movements and pedestrian accidents, whereas for rear-end collisions an increase is sometimes observed. In a recent Israeli study that estimated, inter alia, a safety effect of traffic lights' installation at rural junctions in Israel, the target accident group was also defined as all injury accidents at the treated sites (Hakkert et al, 2002).

Page 159 INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

In the current study, the economic evaluation of safety improvement of a typical rural junction, the target accident group of all injury accidents is considered as well. At the junction considered three injury accidents on average were observed per year. To note, a slightly different consideration of accidents is accepted by the guidelines (Ministry of Transport, 1981), which, for the warrants’ examination, recommend accounting only for accidents associated with vehicle and pedestrian priority problems at the site.

4 Assessment tools

4.1 Method for estimating safety effect

The safety effect from introducing traffic lights (signal control + realignment) at rural junctions in Israel was estimated in a recent study, which was initiated by the Ministry of Transport and conducted by the T&M Company in association with Technion (Hakkert et al, 2002). The study aimed at developing a uniform methodology for evaluating potential safety effects of projects on road infrastructure improvements and estimating safety effects of some 30 types of safety treatments, which were introduced on Israeli roads throughout the 90s. For the estimation of safety effects of road infrastructure improvements, a method combining an after/before comparison with a control group, and with an empirical correction due to selection bias, was proposed. The outline of the method resembles that described in Elvik (1997), whereas in the Israeli study, an extension accounting for changes in traffic volumes was developed. Besides, the reference group statistics, which are necessary for correction of the selection bias, were estimated by the method of sample moments and not on the basis of a regression model. The reference group included sites which are similar to the treatment sites in most engineering characteristics but were left untreated (unchanged) during the “before” periods of all the sites in the treatment group. The demands for the control (comparison) group were as follows: it should be large (to strengthen the significance of the findings), and demonstrate some similarity with the treatment group from the engineering viewpoint. For the treatment type considered, evaluation of the safety effect included three steps: 1) A correction of “before” accident numbers, with the help of reference group statistics, for each site in the treatment group (WP3, 2004 – see Appendix to Chapter 3). 2) An evaluation of the treatment effect at each site by means of the odds-ratio with the comparison group, where for the “before” period the corrected accident numbers (from the first step) are applied. Besides, a correction due to changes in traffic volumes is performed. The formula is:

Page 160 INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

X Estimated effect(θ ) = a δ Ca X m Cb where 1 δ = βc βt Vcb  Vta      Vca  Vtb  where

Xa – the number of accidents observed at the treatment site in the “after” period,

Xm – the corrected number of accidents at the treatment site in the “before” period,

Vta – traffic volume at the treatment site in the “after” period,

Vtb – traffic volume at the treatment site in the “before” period,

Ca – the number of accidents in comparison group sites in the “after” period,

Cb – the number of accidents in comparison group sites in the “before” period,

Vca - traffic volume in comparison group sites in the “after” period,

Vcb - traffic volume in comparison group sites in the “before” period,

βt – the parameter of the safety performance function (a power of relation between traffic volume and the accident number), for treatment sites,

βc – the parameter of safety performance function, for comparison-group sites. 3) Weighting the effects found for separate treatment sites. This is done by means of a standard method known for weighting odds-ratios, where a statistical weight of separate result is defined by the sizes of data sets, which provided this result:

∑ wi ln(θi ) Weighted mean effect(WME) = exp( i ) ∑ wi i 1 1 wi = = VAR(log(θ i )) 1 1 1 1 i + i + i + i X a X b C a C b where

θi - estimate of effect for site i, wi - statistical weight of estimate for site i, i X a – the number of accidents observed at treatment site i, in the “after” period, i X b – the number of accidents at treatment site i, in the “before” period, i C a – the number of accidents in comparison group (for site i), in the “after” period, i C b – the number of accidents in comparison group (for site i), in the “before” period. The 95% confidence interval for the weighed effect is estimated as follows:

Page 161 INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

     z α z α     1−  WME exp 2 ,WME exp 2    w   w    ∑ i   ∑ i    i   i 

The applicable value of the safety effect, i.e. the best estimate of accident reduction associated with the treatment (in percent), is calculated as (1-WME)*100. In the cases of large samples of treatment sites (that diminishes a threat of selection bias and also limits the practical possibility of building a comparable reference group), only steps 2-3 were applied for the evaluation.

4.2 Safety effect of introducing traffic signal control

In the study HAKKERT et al. (2002), the data on the road infrastructure improvements were collected by means of written applications and meetings with the representatives of road and municipal authorities in different country areas. A special database on the issue was established. The data were sought mostly for projects performed in the mid 90s, to have a two-year “before” and two-year “after” period for observation. To represent a specific project in the database, three information elements were defined as crucial: location of the treatment, type of treatment and the period of treatment. For the project to be involved in the evaluation, all three pieces of information had to be thoroughly verified. To provide a minimum but comprehensive presentation of a specific project in the database, a special reporting form was devised which enabled to classify the site and the treatment in accordance with the road layout, area specifics, etc. The data were obtained from the authorities and accomplished by information from detailed maps, field surveys and the publications of the Central Bureau of Statistics (CBS). Within each treatment type for the analysis, a strict definition of the periods “before” and “after” the treatment was provided for each site; a relevant definition of both periods for the comparison-group sites was also attached. The next stage in data preparation was filtering the CBS accident files for the sites and periods required. For each treatment type, files with series of accident numbers were produced for every treatment and comparison group of sites and then processed using the method described in Section 4.1. For the treatment type "introduction of traffic signal control at a rural junction", data were collected on ten projects, which were performed in the north of the country, by the Haifa county of the Public Works Department21. The traffic lights were installed at the junctions over the years 1994-1998. The time period for consideration was 1990-1999, both for the treatment and comparison group sites. For the treatment group, all injury accidents observed at the junctions were considered, whereas for each treated site two-year "before" period and two-year “after” period were separately defined. All injury accidents observed at rural road junctions throughout the country (fitting "before" and "after" periods for each site of treatment) served as a comparison group.

21 Public Works Department (PWD) is the National Road Authority that is responsible for the development and maintenance of the majority of rural roads in Israel.

Page 162 INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

Table 54 details the number of sites (projects) involved in the evaluation, the number of accidents observed at the treatment sites in “before” and “after” periods, the mean value of the safety effect estimated, and the confidence interval for this value. Accident reduction is significant when the whole WME confidence interval is below one. As can be seen from Table 54, a close to significant accident reduction was observed following the treatment: the right boundary of 95% confidence level is slightly over 1. The accident reduction effect of traffic signal control is significant with p=0.11.

Table 54: Safety effect of introducing traffic signals estimated for Israeli conditions Treatment type Estimated WME Number of Number of effect confidence treatment sites accidents at the (WME) interval in the sample treatment sites Introducing signal control at 0.70 (0.453, 1.081) 10 86 rural junctions Source: Hakkert et al, 2002 The average safety effect of introducing traffic signals at rural junctions in Israel was a 30% reduction in injury accidents. This result is comparable with the international value reported by Elvik and Vaa (2004). Accounting for both the significance level and the comparability of finding with the international experience, the above result was classified as “admissible for application” and was recommended for use in evaluations of road infrastructure improvements for Israeli conditions (Hakkert et al, 2002).

4.3 Accident costs

In the current Israeli practice, the average accident cost can be estimated as a sum of injury costs and damage costs of an average accident in the target accident group. The injury costs are a sum of injury-values multiplied by the average number of injuries, with different severity levels, which were observed in the target accident group. The road accident injury values are usually taken as $ 500,000 per fatality, $ 50,000 per serious injury, $ 5,000 per minor injury; the damage value is stated as 15% of the injury costs. Table 55 illustrates the calculation of accident costs for an average injury accident, observed at rural Israeli junctions in 2002. The injury-costs of an average accident are NIS 155,057; with the addition of damage-costs, the value of average injury accident is NIS 178,315 (at 2002 prices). The above values of injury should be treated as conservative because the fatality-value is lower then that estimated accounting for the ‘willingness-to-pay’ approach (MATAT, 2004). Table 55: Estimating costs for an average injury accident at rural junctions in Israel

Value Fatality Serious injury Slight injury Average number of injuries per accident* 0.0275 0.1227 2.571 Injury-values, $ 500,000 50,000 5,000 Total injury-costs of average accident** $ 32,740 or NIS 155,057 Damage costs NIS 23,258 Total costs of an average accident (at 2002 prices) NIS 178,315 *in 2002 **$ 1 = 4.736 NIS (average, in 2002)

Page 163 INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

5 Cost-Benefit Analysis

5.1 General

In this section, a Cost-Benefit Analysis (CBA) of the safety effect from introducing traffic signal control at a rural junction is performed. The CBA compares the measure's safety benefits with the measure's costs, where both values are brought to the same economic framework. As mentioned in Section 1, the main benefits from introducing signal control at a junction come from the improvements of traffic flows, i.e. reduced traffic delays, better use of roads' capacity, etc. Possible safety improvements (an accident reduction following the treatment) present an additional benefit and not the main reason for the application of the measure. In the current practice, a general CBA of introducing signal control at a junction is not obligatory when the warrant for the application of measure is satisfied. In other words, if the traffic volumes at the junction are reasonably high, the traffic lights' installation usually provides apparent economic benefits from the viewpoint of traffic flows. However, a demonstration of these time savings and their costs is not simple as it requires for multiple calculations depending on the traffic signal design parameters, characteristics of traffic flows, approaching speeds, etc. Therefore, in the current evaluation, only benefits associated with safety improvements due to the measure will be estimated and compared with the measure's costs. The evaluation results should be treated as conservative and demonstrating only a part of general benefits associated with the measure. The costs of the measure consist of the initial investment, which is required for the design and introducing signal control at the junction considered, and annual maintenance expenses for providing a proper functioning of the system. Both the costs and benefits are considered for 15 years, with a 7% discount rate (according to the values recommended by the Ministry of Transport – Nohal Prat, 1996); the accumulated discount factor will be 9.108.

5.2 Values of costs and benefits

Introducing traffic signal control at a junction includes both traffic lights' installation and a minor realignment of the junction. The value of the initial investment on the measure should account for the expenses on the traffic signal's design and approval, the junction's redesign and approval, the performance of road paving, building turning lanes, traffic islands and curbs, road signing and marking, and the installation of traffic lights. Typical costs of the measure were estimated by Hakkert et al. (2002) and they amounted to NIS 750,000 (at 2000 prices). At 2002 prices22, the value of initial investment will be NIS 801,525. The annual maintenance expenses present some 5% of the initial investment. Therefore, the total value of costs for the introduction of traffic signal control, over a 15-year period, will be: 801,525 (1 + 0.05* 9.108) = 1,166,539 NIS (at 2002 prices).

22 Change of price index over 2000-2002 is 1.0687.

Page 164 INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

The one-year value of benefits from the expected accident reduction is estimated as a product of the annual number of "before" accidents, the accident reduction factor (the safety effect) and the accident cost. This value is: 3 accidents * 0.3 * 178,315 NIS/ accident= 160,483 NIS (at 2002 prices). The total value of safety benefits from the introduction of traffic signal control, over a 15- year period, will be NIS 1,461,679 (at 2002 prices).

5.3 Cost-Benefit Ratio

Table 56 illustrates the calculation of the cost-benefit ratio (CBR) of the introduction of traffic signal control. The CBR estimated for the measure is 1:1.25. This means that based on safety benefits, only the application of the measure for the rural junction considered appears to be slightly cost-effective. Had the traffic flow benefits been added to the calculations, the CBR would be much higher. Table 56: Calculation of the cost-benefit ratio

Costs Benefits Costs of accidents saved in one year, NIS Initial investment, NIS 801,525 Total benefits in one 160,483 year, NIS Maintenance costs, NIS 40,076 (one-year) Total costs, over 15-year 1,166,539 Total benefits in 15 1,461,679 period, NIS (2002) years, NIS (2002) Total costs, Euro (2002)* 260,388 Total benefits, Euro 326,268 (2002)* Cost-benefit ratio 1 : 1.25 *In 2002: 1 Euro = 4.48 NIS.

6 Decision-Making Process

The cost-benefit analysis of the introduction of traffic signal control at a junction is not common in Israel. Usually, neither safety nor traffic flow benefits are estimated in economic terms. The only estimate that is usually performed is an examination of the site from the viewpoint of warrants for the installation of traffic lights. Both road and local authorities frequently request this measure when any safety problem is identified at the junction. The Ministry of Transport applies efforts to regulate these demands approving the introduction of signal control only for junctions where the measure is really warranted. The estimation of the safety effect from the introduction of signal control is not obligatory according to current guidelines. However, for boundary cases (i.e. when traffic volumes at the junction are slightly lower than the threshold values) such an estimation might provide additional arguments in favour of approving the measure.

Page 165 INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

7 Discussion

In this study, a CBA of a typical example of introducing signal control at a rural junction was considered. The CBA included the safety effect only. A consideration of time savings due to new signal control would strengthen the benefits of the measure. However, such a consideration is complicated and site-specific and, thus, cannot easily be performed within the framework of a mini-CBA. Other possible effects of signalising an intersection are the effects on energy consumption and pollution effects. These were not considered in the present case study. Based on the evaluation of the safety effect only, the measure was found to be beneficial. This is because a certain amount of injury accidents was observed at the junction in the “before” period. However, it is worth mentioning that the economic value of safety benefits is only slightly higher than the costs. The above result alone would not provide a high rank of the site for the measure's application. Obviously, fewer injury accidents in the "before" period would lower the estimated value of benefits, making the results less relevant for the decision-making. The safety effect of introducing signal control, observed under Israeli conditions, was high and close to significant. It was in line with the findings reported by studies in other countries. The CBA presented in this study can be characterized as follows: • the CBA accounts for safety effect only; a consideration of time savings would strengthen the benefits of the measure; • the evaluation findings support the measure's implementation; • to estimate the safety effects, a statistical model was fitted to the accident data from a group of similar sites; the evaluation was in line with the criteria of correct safety evaluation (WP3, 2004); • the accident costs were fitted to the accident type considered, however, they should be treated as conservative as the injury costs do not account for the ‘willingness-to-pay’ component; • the evaluation of the safety effect was initiated by the Ministry of Transport. However, the decision-makers usually do not require a CBA of the measure.

Page 166 INTRODUCING SIGNAL CONTROL AT A RURAL JUNCTION

References

CBS (2003). Road accidents with casualties 2002. Part A: General Summaries. Central Bureau of Statistics, Jerusalem. Elvik, R. (1997). Effects on Accidents of Automatic Speed Enforcement in Norway. Transportation Research Record 1595, TRB, Washington, D. C., pp.14-19. Elvik, R. and Vaa, T. (2004) The Handbook of Road Safety Measures. Elsevier. Hakkert, A.S., Gitelman, V., et al (2002) Development of Method, Guidelines and Tools for Evaluating Safety Effects of Road Infrastructure Improvements. Final report, T&M Company, Ministry of Transport (in Hebrew). MATAT (2004). Road Accidents in Israel: the scope, the characteristics and the estimate of losses to the National Economy. MATAT - Transportation Planning Center Ltd, Ministry of Transport (in Hebrew). Ministry of Transport (1981). Guidelines on design of traffic control signals. The National Transport supervisor, Ministry of Transport (in Hebrew). Nohal Prat (1996). A guideline for economic evaluation of transport projects. 2.0 edition. Ministry of Transport (in Hebrew). WP3 (2004). Improvements in efficiency assessment tools. ROSEBUD.

Page 167 CASE I1: intensification of police enforcement (speed and alcohol)

National Technical University of Athens Department of Transportation Planning and Engineering

ROSEBUD WP4 - CASE I REPORT

INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

BY GEORGE YANNIS AND ELEONORA PAPADIMITRIOU

NTUA / DTPE, GREECE INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

TABLE OF CONTENTS

1 PROBLEM TO SOLVE ...... 171 2 DESCRIPTION...... 171 3 TARGET GROUP ...... 171 4 ASSESSMENT METHOD...... 171 5 ASSESSMENT QUANTIFICATION...... 172 6 ASSESSMENT RESULTS...... 181 7 DECISION MAKING PROCESS...... 181 8 IMPLEMENTATION BARRIERS ...... 181 9 CONCLUSION / DISCUSSION...... 183

Page 169 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

CASE OVERVIEW

Measure Intensification of Speed and Alcohol Enforcement in Greece Problem Road accidents and related casualties presented an increasing trend during the past decade in Greece, mainly due to insufficient maintenance of the road network, inappropriate behaviour of the road users and lack of efficient and systematic enforcement. Since 1998, an important effort was devoted to the improvement of this situation in Greece, focusing on an intensification of enforcement aimed at improving driver behaviour. Target Group Drivers, mainly on the interurban road network Targets a) Increase in the number of police controls for speeding and drinking-and-driving b) Decrease in the number of road accidents and related casualties Initiator National Police Decision-makers National Police Costs Police Labour Costs, Police Vehicle Costs, Police Equipment Costs Benefits Fatal and injury accidents prevented Cost-Benefit Ratio 1:6.6 to 1:9.7

Page 170 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

1 Problem

Road accidents and related casualties increased during the past decade in Greece, mainly due to insufficient maintenance of the road network, inappropriate behaviour of the road users and lack of efficient and systematic enforcement [NTUA/DTPE, 2003]. Since 1998, an important effort was devoted to the improvement of this situation in Greece, focusing on an intensification of enforcement aimed at improving driver behaviour.

2 Description

In 1998, the Greek Traffic Police started the intensification of road safety enforcement, having set as the target the gradual increase of road controls for the two most important infringements: speeding and drinking & driving. Since then, all controls and related infringements recorded are systematically monitored and the related enforcement and casualty results at the local and national level are regularly published, as shown in the following table with basic road safety related trends in Greece. Seat belt and helmet use were two additional offences, which the police started to enforce more systematically in 2002. Table 57: Basic road safety and enforcement trends in Greece (1998-2002)

1998 1999 2000 2001 2002 5-year change Injury road accidents 24,819 24,231 23,127 19,710 16,852 -32% Fatalities 2,182 2,116 2,088 1,895 1,654 -24% Vehicle fleet (x1000) 4,323 4,690 5,061 5,390 5,741 33% Speed infringements 92,122 97,947 175,075 316,451 418,421 354% Drinking & driving infringements 13,996 17,665 30,507 49,464 48,947 250% Drinking & driving controls 202,161 246,611 365,388 710,998 1,034,502 412%

3 Target Group

The target group of the measure included the entire population of Greek drivers. Although the intensification of enforcement was more significant on the interurban road network, it is considered that the entire number of accidents was affected. In particular, the enforcement was nationwide and concerned all types of traffic violations. Moreover, previous research allows for the quantification of the particular effect of speed and alcohol enforcement in particular regions of Greece, as described in the following sections.

4 Assessment method

The present research concerns a cost-benefit evaluation of police enforcement for speeding and drinking & driving in Greece for the period 1998-2002. The evaluation was based on detailed police controls and infringements data, available by the police for the examined period. Additional information was collected by means of interviews with police officers in order to estimate the implementation costs of the measures. As far as safety benefits are concerned, the results of three recent studies were used; one study concerned the calculation of accident economic costs in Greece, one study on the

Page 171 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

‘willingness-to-pay’ for accident risk reduction in Greece, and one study concerned the quantification of the safety effect of enforcement and other safety related parameters in Greece. It should be noted that enforcement costs include police labour, vehicle and equipment costs, whereas enforcement benefits exclusively refer to safety effects.

5 Assessment Quantification

5.1 Enforcement Costs

Enforcement costs include police labour costs, police vehicle costs and police speed and alcohol enforcement equipment costs (speed cameras, alcoholmeters etc.). As the intensification of enforcement in the examined period was not foreseen as part of a specific project with a specific budget and resource allocation, there was very little information available on police-related costs. The additional necessary information for CBA calculations was obtained by means of exhaustive interviews with Head Officers of the police. In particular, on the basis of the available detailed information on yearly numbers of speed and alcohol infringements, the interviews tried to yield the related labour and capital parameters through the adoption of typical conversion measures.

5.1.1 Police Labour Costs

Table 58: Police Labour Costs for Speed Enforcement in Greece (1998-2002)*

1999 2000 2001 2002 Number of infringements 97,947 175,075 316,451 418,421 typical days 73,460 131,306 237,338 313,816 special days 24,487 43,769 79,113 104,605 Number of shifts typical days 4,897 8,754 15,823 20,921 special days 1,224 2,188 3,956 5,230 Shifts Labour Persons 3 3 3 3 Person-hours 8 8 8 8 Hourly rate (€) 7.5 7.5 7.5 7.5 Shifts Costs (€) 1,101,904 1,969,594 3,560,074 4,707,236 Number of prosecutions 2,938 5,252 9,494 12,553 Prosecution Police Labour Persons 1 1 1 1 Person-hours 14 14 14 14 Hourly rate (€) 7.5 7.5 7.5 7.5 Prosecution Costs (€) 308,533 551,486 996,821 1,318,026 Total Labour Costs (€) 1,410,437 2,521,080 4,556,894 6,025,262 *prices of 2002

Page 172 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

The total yearly labour cost of speed enforcement is summarized in the table above. The calculations for speed enforcement are based on the following assumptions, as reported by Head Police Officers interviewed, based on experience: • 75% of speed infringements are recorded on typical days • 25% of speed infringements are recorded on special days (weekends, holidays, special events) • An average of 15 speed infringements per shift are recorded on typical days • An average of 20 speed infringements per shift are recorded on special days • 3% of speed infringements recorded result to driver's prosecution, both on typical and special days As far as alcohol enforcement labour is concerned, the calculations are presented in the following table.

Table 59: Police Labour Costs for Alcohol Enforcement in Greece (1998-2002)*

1999 2000 2001 2002 Number of infringements 17,665 30,507 49,464 48,947 typical days 13,249 22,880 37,098 36,710 special days 4,416 7,627 12,366 12,237 Number of shifts typical days 13,249 22,880 37,098 36,710 special days 2,208 3,813 6,183 6,118 Shifts Labour Persons 3 3 3 3 Person-hours 8 8 8 8 Hourly rate (€) 7.5 7.5 7.5 7.5 Shifts Costs (€) 2,782,238 4,804,853 7,790,580 7,709,153 Number of prosecutions 1,767 3,051 4,946 4,895 Prosecutions Labour Persons 1 1 1 1 Person-hours 14 14 14 14 Hourly rate (€) 7.5 7.5 7.5 7.5 Prosecutions Costs (€) 185,483 320,324 519,372 513,944 Total Labour Costs (€) 2,967,720 5,125,176 8,309,952 8,223,096 *prices of 2002

The respective assumptions for alcohol enforcement are the following: • 75% of alcohol infringements are recorded on typical days • 25% of alcohol infringements are recorded on special days • An average of 1 alcohol infringements per shift is recorded on typical days • An average of 2 alcohol infringements per shift are recorded on special days • 10% of alcohol infringements recorded result to driver's prosecution, both on typical and special days

Page 173 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

Based on the information above, the yearly numbers of police control shifts on speed and alcohol enforcement and prosecutions for speeding and drinking & driving were calculated. Additionally, a detailed labour breakdown for control shifts and prosecutions obtained though interviews (number of persons and person-hours of a typical control shift/prosecution, typical policeman hourly rate) was used to calculate the total yearly labour costs for alcohol enforcement. It should be noted that the police person-hour rate (€) refers to year 2002. In particular: • 3 policemen are involved in one control shift for 8 hours each • 1 policeman is involved in an prosecution for a total of 14 hours • The hourly rate of a policeman is 7.5 €

5.1.2 Police Vehicle Costs

The calculation of vehicle costs is based on the number of police control shifts and prosecutions, which was calculated as described above on the basis of the interviews. Additional information on the use of police vehicles collected during the interviews was also exploited. The results are summarized in the following tables.

Table 60: Police Vehicle Costs for Speed Enforcement in Greece (1998-2002)*

1999 2000 2001 2002 Number of shifts 6,122 10,942 19,778 26,151 Number of prosecutions 2,938 5,252 9,494 12,553 Shifts Vehicle costs Number of vehicles 1 1 1 1 Average distance 20 20 20 20 Unit Cost per Km (€) 0.1 0.1 0.1 0.1 Vehicle Cost (€) 12,243 21,884 39,556 52,303 Prosecutions Vehicle Costs Number of vehicles 1 1 1 1 Average distance5555 Unit Cost per Km (€) 0.1 0.1 0.1 0.1 Vehicle Cost (€) 1,469 2,626 4,747 6,276 Total Vehicle Costs (€) 13,713 24,511 44,303 58,579 *prices of 2002 As far as speed enforcement is concerned, the following assumptions were included: • 1 police vehicle is used in each shift • 1 police vehicle is used for each driver's prosecution • The average total distance travelled for each shift is 20 km • The average total distance travelled for each prosecution is 5 km

Page 174 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

Table 61: Police Vehicle Costs for Alcohol Enforcement in Greece (1998-2002)*

1999 2000 2001 2002 Number of shifts 15,457 26,694 43,281 42,829 Number of prosecutions 2,938 5,252 9,494 12,553 Shifts Vehicle costs Number of vehicles 1 1 1 1 Average distance5555 Unit Cost per Km (€) 0.1 0.1 0.1 0.1 Vehicle Cost (€) 7,728 13,347 21,641 21,414 Prosecutions Vehicle Costs Number of vehicles 1 1 1 1 Average distance5555 Unit Cost per Km (€) 0.1 0.1 0.1 0.1 Vehicle Cost (€) 1,469 2,626 4,747 6,276 Total Vehicle Costs (€) 9,198 15,973 26,387 27,691 *prices of 2002 As far as alcohol enforcement is concerned, the following assumptions were included: • 1 police vehicle is used in each shift • 1 police vehicle is used for each driver's prosecution • The average total distance travelled for each shift is 5 km • The average total distance travelled for each prosecution is 5 km Additionally, the average police vehicle cost per kilometre was considered equal to 0.10 €/km (referring to year 2002) according to a recent study on accident costs in Greece [LIAKOPOULOS, 2002].

5.1.3 Police Equipment Costs

The number of available devices used for speed and alcohol enforcement for the year 2002 was obtained from the Technical Services of the Police. However, no information on the respective numbers for the year 1998 was available. According to the information collected during the interviews, a reasonable assumption would be to consider that the enforcement equipment was doubled in the examined period. Table 62: Police Equipment Costs for Speed and Alcohol Enforcement in Greece (1998-2002)*

1998 2002 Number of portable speed guns 231 462 Unit cost (€) 600.00 Number of in-car radars 31 62 Unit cost (€) 500.00 Number of speed guns on tripod 20 39 Unit cost (€) 300.00 Number of alcoholmeters 467 934 Unit cost (€) 10.00 Total Equipment Costs (€) 164,620

Page 175 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

*prices of 2002

5.2 Enforcement Benefits

In the framework of this study, the benefits examined exclusively concern safety benefits, as no significant social or environmental costs were expected from the intensification of speed and alcohol enforcement. The available results of previous research allowed for the direct calculation of the number of accidents prevented by the measures, as described in detail in the following sections.

5.2.1 Number of accidents prevented

For the estimation of the number of accidents prevented from the intensification of speed and alcohol enforcement, the results of a recent research study were used [AGAPAKIS, MYGIAKI, 2003]. This research concerned a macroscopic investigation of the effect of enforcement on road safety improvement in Greece aimed in particular at determining the separate effect of different types of enforcement (speeding, drinking and driving, violating signals, failing to yield etc.), as well as the effect of other safety related parameters (vehicles fleet, vehicle ownership, population) on the significant overall improvement of road safety in Greece during the last few years. This study included two distinct parts; the first part concerned a cluster analysis aimed at identifying groups with similar characteristics within the 52 departments of Greece. In particular, road network, population density, vehicle ownership, traffic infringements and accidents characteristics were used for the separation of Greece in four groups of departments, as follows: Figure 18: Clustering of the departments of Greece in groups of similar accident and infringement rates

Group I

Group II

Group III

Group IV

• Group I included the Athens and Thessaloniki large urban regions, which present high accident and infringement rates • Group II included 5 large departments with relatively high population density and accident and infringement frequencies

Page 176 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

• Group III included 22 departments with relatively medium population density, high accident frequencies and medium infringement frequencies • Group IV included 22 smaller departments with relatively low population density, accident and infringement frequencies The second part of the study concerned the development of Poisson regression models for the quantification of the separate effect of various types of enforcement, as well as other parameters on the total number of accidents in each group of departments. In each case, the marginal effects of the various significant parameters were also calculated. Additionally, the modelling process was developed for two different assumptions concerning the effect of enforcement, resulting in two categories of models: • Models with no time-halo in the effect of enforcement • Models with a time-halo in the effect of enforcement The above classification rises from the international experience, according to which there may be a delay of several weeks before a significant effect of enforcement is observed (Holland, Corner, 1996, Vaa, 1997). This "time halo effect" was examined in the framework of the analysis of intensification of enforcement in Greece. It is interesting to note that, among the various types of enforcement examined in this study, the enforcement of speeding and drinking & driving was found to have a significant effect on the total number of accidents only in Groups II and IV, whereas in the other groups, other types of enforcement were found significant, such as traffic signals violations, failing to yield etc. The quantified effects are presented in detail below.

5.2.2 Consideration with no time-halo in the effects of enforcement

The first group of models is based on the assumption that there is no time-halo (delay) in the effect of enforcement on the total number of road accidents. In this scenario, the number of police controls and infringements of a certain period is considered to directly affect the number of accidents of this period. As mentioned above, the effect of speed and alcohol enforcement was significant in Groups of departments II and IV. In particular, it was found that an increase of 1000 speed infringements prevents approximately one accident in Group II departments and two accidents in Group IV departments. Additionally, it was found that an increase of 1000 alcohol controls prevents approximately two accidents in Group II departments and 1 accident in Group IV departments. In the framework of the present research, the above results were combined with the related enforcement trends data for 1998-2002, which is available in detail from the National Police, in order to calculate the total number of accidents prevented from the intensification of enforcement in the examined period. The results are presented in detail in the following Table 63. According to the results of the consideration without delay in the effects of enforcement, a total number of 772 accidents were prevented in the examined period in Greece. This consideration will be adopted as the "conservative scenario" of the present cost-benefit evaluation, corresponding to a minimum effect of enforcement.

Page 177 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

Table 63: Marginal effects of enforcement and number of accidents prevented (1999-2002), no time-halo-effect

Department Group I II III IV Marginal effects Speed infringements -1.239 -1.542 Alcohol controls -1.929 -1.373 Speed infringements 1998 23,867 9,579 42,028 14,648 1999 32,480 16,091 49,169 19,899 2000 37,324 31,533 74,323 30,112 2001 68,397 64,966 128,924 54,164 2002 105,025 82,531 161,297 69,568 Alcohol controls 1998 100,955 13,584 62,655 24,967 1999 104,540 19,485 87,415 35,171 2000 128,287 54,498 121,775 60,828 2001 211,273 151,943 235,716 112,066 2002 290,052 213,138 351,888 179,552 Accidents prevented 1999 0 19 0 22 2000 0 87 0 51 2001 0 229 0 107 2002 0 140 0 116 TOTAL 772

5.2.3 Consideration of a two-month time-halo in the effect of enforcement

The second group of models was based on the assumption that there is a two-month time- halo (delay) in the effect of enforcement on the total number of road accidents. More specifically, in these models, the number of controls and infringements of one month were combined with the accidents of the next third month. As mentioned above, the effect of speed and alcohol enforcement was significant in Groups of departments II and IV. In particular, it was found that an increase of 1000 speed infringements prevents approximately one accident in Group II departments and two accidents in Group IV departments. Additionally, it was found that an increase of 1000 alcohol controls prevents approximately two accidents in Group II departments and one accident in Group IV departments. Accordingly, the results were combined with the related enforcement trends data for 1998- 2002 in order to calculate the total number of accidents prevented from the intensification of enforcement in the examined period. The results are presented in detail in the following Table 64.

Page 178 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

Table 64: Marginal effects of enforcement and number of accidents prevented (1999-2002) - Two-months time-halo-effect

Department Group I II III IV Marginal effects Speed infringements -2.224 -2.053 Alcohol controls -2.265 -2.684 Accidents prevented 1999 0 28 0 38 2000 0 114 0 90 2001 0 295 0 187 2002 0 178 0 213 TOTAL 1,142

The results of the consideration with two-month time-halo in the effects of enforcement indicate a total number of 1,142 accidents prevented in the examined period in Greece. This consideration will be adopted as the "best scenario" of the present cost-benefit evaluation, corresponding to a maximum effect of enforcement.

5.2.4 Accident costs

The estimation of average accident costs was carried out on the basis of a recent study on accident costs in Greece (LIAKOPOULOS, 2002). This study concerned the estimation of the costs of various components of accidents (material damage costs, generalized costs, human costs) for fatal accidents, injury accidents and material damage accidents, including: • Material damage costs • Police costs • Fire brigade costs • Insurance companies costs • Court costs • Lost production output • Pain and grief • Rehabilitation costs • Hospital treatment costs • First aid and transportation costs The various costs were calculated by means of an exhaustive data collection process addressed to various organizations (e.g. National Statistical Service of Greece, National Police, Fire Service of Greece, Emergency Medical Service of Greece, hospitals, courts, insurance companies). Additional parameters were adopted on the basis of estimations from experts in each field, as well as the existing international literature. It should be noted, however, that the above study did not adequately account for the human cost component, as the pain and grief parameters as reported in the courts are not sufficiently representative of the human cost. For that purpose, a separate investigation for human cost in Greece was carried out in the framework of present research. In particular, human cost was estimated according to the following formula: Page 179 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

VoSL = (NAEIS) / (LSE) Where: VoSL: Value of Statistical Life NAEIS: National Annual Expenditure on Improving Safety LSE: Expected Lives Saved from this Expenditure Annually

In particular, the calculations included parameters such as the percentage of family annual income that each person is willing to pay in his/her entire life in order to reduce the probability of accident involvement of himself/herself or of any family person by 50%, the average members per family in Greece, the proportion of families with an economically active member, the average family annual income in Greece, the national population, the life expectancy in Greece and the current and new accident risk. In regards the percentage of family annual income that each person is willing to pay in his/her entire life in order to reduce the probability of accident involvement by 50%, the results of a recent "willingness-to-pay" survey were used [AGGELOUSI, KANNELOPOULOU, 2002]. In this survey, respondents were asked the percentage of annual income they were willing to pay to reduce the probability of fatal accident, injury accident and material damage accident involvement by 50%. It should be noted that in the willingness-to-pay survey, respondents were also asked to rate various types of accidents and injuries in order to identify their perception on injury severity. On the basis of the results in the present research, the value corresponding to injury accidents is considered to adequately represent serious injury accidents, whereas the value for material damage accidents is considered to adequately represent both slight injury and material damage accidents. On the basis of the above, the human cost of accidents in Greece was estimated as: VoSL = 612,140.72 €/person for fatal accidents VoSL = 467,703.02 €/person for serious injury accidents VoSL = 206,339.57 €/person for minor injury and material damage accidents It should also be underlined that the calculations concern prices of 1999. In order to calculate the average accident cost in Greece, the costs of fatal and injury accidents were weighted in relation to the average distribution of accident casualties per casualty severity in Greece. In the following Table 65, the parameters concerning accident costs in Greece are summarized on the basis of the previous research used and the additional calculations carried out: Table 65: Calculation of average accident cost in Greece (1999)

Cost of Accidents with Fatalities Seriously Injured Slightly Injured Material Damage cost 28,769.42 18,174.91 13,904.19 Generalised cost 442,466.54 23,906.66 6,960.30 Human cost 612,140.72 467,703.02 206,339.57 Total cost 1,083,376.68 509,784.59 227,204.06 Proportion of casualties 5.81% 11.60% 82.59% Average accident cost 309,723.25

Page 180 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

6 Assessment Results

On the basis of the detailed approach described in the previous sections, the cost-benefit ratio was calculated for the "conservative" scenario and the "best" scenario. An accumulated discount factor was applied to the benefits calculation on the basis of an interest rate of 4% (National Statistical Service of Greece, 2003). Table 66: Calculation of the cost/benefit ratio for the "conservative" scenario (1999-2002)

Present value of benefits 1999 2000 2001 2002 Number of accidents prevented 42 138 337 256 Average accident cost (€) 309,723.25 309,723.25 309,723.25 309,723.25 Accumulated discount factor 1.000 1.040 1.082 1.125 Total (€) 12,871,643.25 44,338,168.72 112,837,881.79 89,265,963.13 Present value of costs Cost of Speed Enforcement (€) 1,424,149.38 2,545,590.50 4,601,197.54 6,243,791.34 Cost of Alcohol Enforcement (€) 2,976,917.64 5,141,148.94 8,336,339.27 8,255,456.63 Benefit/Cost Ratio 6.6:1

As shown in the above Table 66, the "conservative" scenario yielded a very high benefit- cost ratio equal to (6.6:1). In particular, the total value of benefits for this scenario were calculated equal to 274,696,321.34 €, whereas the enforcement implementation costs totalled 39,524,591.23 €, all values referring to year 2002.

Table 67: Calculation of the cost/benefit ratio for the "best" scenario (1999-2002)

Present value of benefits 1999 2000 2001 2002 Number of accidents prevented 66 203 482 390 Average accident cost (€) 309,723.25 309,723.25 309,723.25 309,723.25 Accumulated discount factor 1.000 1.040 1.082 1.125 Total 20,446,780.45 65,542,783.83 161,458,163.95 136,023,786.23 Present value of costs Cost of Speed Enforcement (€) 1,424,149.38 2,545,590.50 4,601,197.54 6,243,791.34 Cost of Alcohol Enforcement (€) 2,976,917.64 5,141,148.94 8,336,339.27 8,255,456.63 Benefit/Cost Ratio 9.7:1

Accordingly, as shown in the above Table 67, the "best" scenario yielded an even higher benefit-cost ratio equal to (9.7:1). In particular, the total 1999-2002 value of benefits for this scenario were found equal to 406,219,308.40 €, whereas the 1999-2002 enforcement implementation cost totalled 39,524,591.23 €, all values referring to year 2002. In both scenarios, the nationwide intensification of speed and alcohol enforcement in Greece was found to be highly cost-effective.

7 Decision-Making Process

The results of this research were presented to Head Officers of the police at the Ministry of Public Order. Although these decision-makers were not familiar with efficiency assessment

Page 181 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL) in terms of cost-benefit analyses, they responded positively towards this work from the first stages, contributed with data and other available information, and were very helpful in dealing with lack of data when necessary. Furthermore, decision-makers were very interested in the results. The high benefit-cost ratios were received as a confirmation of the important role of the police in road safety and a validation of the systematic efforts of the police to contribute in the reduction of road accidents and casualties. Consequently, they intend to communicate these results to their superiors, to the press, as well as the Head Police Officers of the various regional police departments. They were also asked on their possible response if the results were negative or less encouraging. They replied that they would try to identify the more cost-effective cases among the results and focus their efforts accordingly. Decision-makers also expressed a high interest for more analyses and results concerning other types of police enforcement and other road safety related activities of the police. They also underlined that these results would have been even more useful if they were available at earlier stages of the intensification of enforcement.

8 Implementation barriers

• As far as the implementation of the measures is concerned, the basic barrier concerned the inefficiency of the process for the payment of the infringement ticket, as several different authorities are involved in the process (police, municipalities, tax computer centre, etc.). Other related barriers concerned the lack of appropriate number of policemen and the reactions from the drivers and the policemen against the systematic controls (using several different pretexts). These parameters were the main difficulties encountered during the early implementation period. • As far as the present evaluation is concerned, the main difficulty concerned the lack of detailed and accurate data on the specific resources allocated in the intensification of enforcement. As in most countries (ESCAPE, 2003), no standards to measure police intensity existed in Greece and no system of performance indicators for enforcement activity was developed. Additionally, neither police headquarters nor road safety authorities use such performance indicators. Consequently, the systematic recording of the number of police controls and related infringements achieved during the 1998-2002 period in Greece contributed important progress in the monitoring of the enforcement activity as well as the road safety level in Greece. • Additionally, no systematic and official cost data are available in Greece. In particular, police costs are not systematically recorded in relation to specific actions, as labour and capital allocation is optimised according to the specific needs of each circumstance. As far as accident costs are concerned, no social values of reference are officially published, and the estimated values are based on survey results. • In the present evaluation, the lack of appropriate data for cost-benefit evaluation purposes was overcome by means of exhaustive interviews with experienced Head Officers of the police who had also been actively involved in both the decision-making Page 182 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

process and the monitoring of police effort. Additionally, existing research in Greece was further used to yield the necessary parameters for the computation of cost-benefit ratios.

9 Conclusion / Discussion

There is certainly a correlation between systematic road safety enforcement and the number of road accidents. This road safety enforcement intensification is one of the two basic reasons (the other one is congestion) that may explain the important decrease observed in the number of road accidents, persons killed and injured during the last five years in Greece. Previous research on enforcement assessment has indicated that only a significant increase in enforcement level may affect the number of accidents [BJØRNSKAU, ELVIK, 2003]. Additionally, very little validation of enforcement effect at the national level has been available in international literature. In particular, most evaluation attempts concern a temporary increase in local resources or concentrated enforcement efforts in a selected area [ESCAPE, 2003]. However, as far as Greece is concerned, the measures were implemented at the national level, and a systematic intensification of enforcement covering all types of violations was achieved. The present research has revealed a limited exploitation of assessment methods in the decision-making process in Greece. This phenomenon is not specifically related to the processes and administrations related to the particular research on enforcement, as CBA and CEA evaluations are not commonly used in general in Greece. As far as the particular case is concerned, the lack of systematic and appropriate cost data complicated the assessment process. The co-operation of the decision-makers who provided useful data based on their experience was very important in dealing with this problem. However, it is obvious that a lot of additional effort is required in order to achieve a systematic recording of police labour and capital costs, in a similar way that the related controls and infringements were monitored since the intensification of police enforcement in Greece. However, the important benefit obtained from the intensification of speed and alcohol enforcement in terms of number of accidents and casualties prevented could motivate decision-makers towards further improvement of the implementation and monitoring of the measures. Additionally, it is obvious that decision-makers respond very positively to results of CBA and CEA evaluations when these are available, as their efforts and policies are confirmed. Nevertheless, such efficiency assessment is rarely initiated, especially at the national level.

Page 183 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

REFERENCES

AGAPAKIS, J., MYGIAKI, E., (2003): "Macroscopic investigation of the effect of enforcement on the improvement of road safety on Greece", Diploma Thesis, NTUA, School of Civil Engineering, Department of Transportation Planning and Engineering, Athens. BJØRNSKAU, T., ELVIK, R., (1992): "Can road traffic law enforcement permanently reduce the number of accidents?", Accident Analysis & Prevention, Volume 24, Issue 5, Pages 507-520. ESCAPE CONSORTIUM (2003): "Traffic enforcement in Europe: Effects, measures, needs and future", The “Escape” Project Final report, Contract Nº: RO-98-RS.3047, 4th RTD Framework Programme. HOLLAND, C.A., CONNER, M.T., (1996): "Exceeding the speed limit: An evaluation of the effectiveness of a police intervention", Accident Analysis & Prevention, Volume 28, Issue 5, Pages 587-597. KANELLOPOULOU, A., AGGELOUSSI, K., (2002): "Estimation of the human cost of road accidents and drivers' sensitivity towards accident risk - A willingness-to-pay technique and a stated-preference technique", Diploma Thesis, NTUA, School of Civil Engineering, Department of Transportation Planning and Engineering, Athens. LIAKOPOULOS, D. (2002): "Development of a model for the estimation of the economic benefits from accident reduction in Greece", Diploma Thesis, NTUA, School of Civil Engineering, Department of Transportation Planning and Engineering, Athens. NATIONAL STATISTICAL SERVICE OF GREECE, (2003): "Greece in figures", Official Publication of the National Statistical Service of Greece, Athens (www.statistics.gr). National Technical University of Athens, Dept. of Transportation Planning and Engineering, (2003): "A strategic plan for the improvement of road safety in Greece 1998-2002", Ministry of Economy and Finance. TSAMBOULAS, D., (2004): "Evaluation of transport infrastructure projects", NTUA, School of Civil Engineering, Department of Transportation Planning and Engineering, Athens. VAA, T., (1997): "Increased police enforcement: Effects on speed", Accident Analysis & Prevention, Volume 29, Issue 3, Pages 373-385. YANNIS, G., KANELLOPOULOU, A., AGGELOUSSI, K., TSAMBOULAS, D., (2003): "Modelling driver choices towards accident risk reduction", Article In Press, Safety Science.

Page 184 CASE I2: Concentrated General enforcement on interurban roads in israel

Technion - Israel Institute of Technology Transportation Research Institute

ROSEBUD WP4 - CASE I REPORT

CONCENTRATED GENERAL ENFORCEMENT ON INTERURBAN ROADS IN ISRAEL

BY VICTORIA GITELMAN AND SHALOM HAKKERT,

TRANSPORTATION RESEARCH INSTITUTE, TECHNION, ISRAEL INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

TABLE OF CONTENTS

1 PROBLEM ...... 188 2 DESCRIPTION OF MEASURE...... 189 2.1 The enforcement project ...... 189 2.2 The follow-up study ...... 190 3 TARGET ACCIDENT GROUP...... 191 4 ASSESSMENT RESULTS...... 191 4.1 Monitoring of police activity ...... 191 4.2 Accident analysis ...... 193 5 COST-BENEFIT ANALYSIS...... 195 5.1 General ...... 195 5.2 Costs...... 196 5.3 Benefits ...... 197 5.4 Computation of the Cost-Benefit Ratio...... 199 6 DECISION-MAKING PROCESS...... 199 7 ROLE OF BARRIERS ...... 200 8 DISCUSSION...... 200

Page 186 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

CASE OVERVIEW

Measure A police project of concentrated general enforcement on interurban roads Problem Reducing the high numbers of severe accidents on main rural roads Target Group Accidents of all types, with fatalities or serious injuries Targets Improving drivers' behaviour and diminishing severe accidents on main rural roads Initiator National Road Safety Authority and the Police Traffic Department Decision-makers National Road Safety Authority and the Police Command Costs Additional personnel costs, additional vehicle fleet expenses and the costs of a publicity campaign that accompanied the police project; paid by the National Road Safety Authority and the Ministry of Interior Security Benefits The benefits stemmed from prevention of severe accidents, which were attained during the project's performance. The driving public and the national economy will benefit. Cost-Benefit Ratio Ranges from 1:3.5 for a "conservative estimate" of the accidents prevented to 1:5 for the "best estimate".

Page 187 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

1 Problem

Traffic law enforcement is believed to be a factor that contributes significantly to normative road user behaviour and road safety. Traffic rules are usually enforced by traffic police forces whose activity and success are generally limited by the resources that can be applied and by established priorities. Numerous long and short-term enforcement projects frequently accompanied by research studies have been performed throughout the world over the last forty years, aimed at increasing the effects of police activity on driver behaviour and road safety and, concurrently, to improve the enforcement methods in use. Examples of literature surveys that summarize the findings of this research are Fitzpatrick (1992); Bjornskau and Elvik (1992); Zaal (1994); Oei (1998); and OECD (1999). A number of large-scale European studies on the subject funded by the European Commission have also been conducted over the past ten years, including GADGET, ESCAPE and VERA. Over the last two decades, traffic rules' enforcement became a significant share of activity of the Israeli police. The National Traffic Police (NTP) in Israel was established in 1991 as an operational branch of the national police, when all existing interurban traffic units came under its direct command. At the beginning of 1997 the NTP’s responsibility covered over 3100 kilometres of interurban roads, where the traffic police forces counted more than 400 patrol officers, about 150 patrol vehicles and about 70 units of mobile enforcement tools (speed guns and photo radar cameras). The NTP has always been looking for more effective forms for deployment of its forces. The approach usually applied in Israel for deployment of patrol cars on road sections can be termed “correlative”, whereby the number of accidents that occurred on a certain road and the traffic volumes determine the road’s priority for police enforcement. (A detailed description of the method is given in Hakkert et al, 1991). This approach is common for the annual and other typical plans of activity of the traffic police. Besides, as the NTP bears the responsibility for the whole network of interurban roads, during the 90s two nationwide enforcement experiments took place. The first of these enforcement projects was performed following the NTP foundation in 1991, and lasted for 21 months (Zaidel et al, 1994). At the beginning of 1997, the NTP, with the support of the National Road Safety Authority, undertook a redeployment of its forces and started the second nationwide enforcement project. This was called the 700-project as its basic idea was to concentrate the major part of the NTP forces on about 700 kilometres (some 20%) of interurban roads which in 1996 contained the majority of all interurban accidents and about half of all severe rural accident locations. The project began in April 1997 and lasted for one year. The new deployment was aimed at increasing the enforcement activity on the roads under focus, to properly combine the traffic and safety issues in everyday police operations, and to lead the traffic police to a more effective use of its resources. In contrast to most reported projects in this field, which usually tend to be localized and/or focus on specific target behaviours and populations, the 700-project was planned on a wide geographical scale with the intention to improve the general functioning of the NTP forces and to determine the resource allocation and field activity modes, providing maximal influence on drivers’ behaviour and road safety. From reviews on the subject (e.g. Bjornskau and Elvik, 1992) it follows that changes in drivers’ behaviour and a decrease in accident frequency can be expected when the enforcement intensity has been increased by at least a factor of three. Due to its size, the 700-project could not satisfy this demand. However, accounting for a general deterrence effect of such wide-scale enforcement, one might expect changes in drivers' behaviour and in the end, in the accidents.

Page 188 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

Besides, the potential to influence driver behaviour and accident occurrences might increase when conventional enforcement is fortified by non-trivial enforcement tactics such as a random scheduled deployment or the use of the field experience of the police officers when the activity types are selected.

2 Description of measure

2.1 The enforcement project

The "700-project" was the second large-scale experiment of the National Traffic Police. The project began in April 1997 and lasted for one year. The project involved ten out of thirteen regional police subdivisions that comprised more than 90% of the NTP staff. Within the project, the NTP declared a redeployment of its forces, concentrating the major part of its resources on about 700 km of interurban roads. These were the most heavily travelled roads throughout the country which, in 1996, contained some 60% of all interurban accidents and about half of the severe accident locations (severe accidents are those with serious casualties or fatalities). The declared purpose of the project was to achieve a reduction in severe accidents on the roads in focus. The “700 project” roads included fifteen road sections (Table 67) with a traffic volume of 17-80 thousand vehicles per day. Four of the roads, i.e. roads No 65, 70, “4-center” and “40-south” were declared as the highest priority roads, intended for maximum enforcement "coverage". The police planned three-shift patrols everyday on the highest priority roads, two working day shifts on other project's roads, and 2-3 shifts per week for the rest (of the rural network). The enforcement was announced to emphasize severe violations (speeding, not keeping to the right, non-compliance with traffic signs and other moving violations), however in practice it was not limited to severe violations only. Being inspired by the Australian experience of enforcement programs combined with publicity campaigns (Cameron et al, 1996), the Israeli Road Safety Authority initiated a publicity campaign, which was launched simultaneously with the beginning of the 700- project. This was the first experiment in NTP history where publicity accompanied the police operation in a controlled manner. The campaign consisted of TV and radio advertisements, press announcements, outdoor advertising and special yellow sign posts indicating intensive enforcement that were erected on the shoulders of the project roads. Two purposes were determined for the campaign: a) to inform the public about the police enforcement project, its territory and major violations enforced; and b) to strengthen the public feeling that the risk of apprehension had grown for those who violated traffic rules on the project roads. The publicity campaign in the media lasted four months, from April to July 1997. The static outdoor advertising and signposts were left in the field till the end of the police project.

Page 189 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

Table 68: 700-project road sections

Region Road No Section: from km Length, km AADT, 1000 vehicles Road Type* to km North 65 0-60 60 30.2 D 70 10-52.9 42.9 25.1 D 2 55-100 45 39.7 D 4 157-200 43 18.8 S 85 1-31 30 21.4 D+S** 79 0-27 27 17.3 S 77 49-75.7 26.7 18.0 S 75 20-49 29 19.4 D Centre 2 28-55 27 79.5 D 4 85-157 72 62.3 D+F** 40 248-301 53 35.3 D 44 10-35 25 29.1 D 1 4-56 52 47.7 F South 4 51-85 34 27.2 D 40 189-248 59 30.9 D *F – Freeway; D - Dual-carriageway; S - Single-carriageway **Includes sections of both types

2.2 The follow-up study

An assessment study was conducted for the purpose of the follow-up of the actual project performance and of the project’s influence on drivers’ behaviour and on road safety. This was performed by Technion – the Transportation Research Institute in co-operation with the Technion Statistics Laboratory. The assessment project began in March 1997 and accompanied the police activity for the whole year [HAKKERT et al, 1998]. The underlying rationale of the evaluation study was based on the assumption of a chain of relations between police activities and road safety [e.g. OECD, 1999; HAKKERT et al., 2001]. It is hypothesized that the new deployment of police forces will lead to increased enforcement on the project roads; the latter implies a growth in the actual risk of being detected for traffic rules’ violations which, together with the accompanying publicity, raises the subjective probability of apprehension perceived by the drivers. The subjective probability of apprehension, together with the expected punishment for violators that is meted out by the judicial process, constitutes the deterrence effect. Ultimately, the deterrence and the detection in combination with proper education and training may cause positive changes in driving norms and actual traffic behaviour in a manner that would manifest itself in a reduction in accidents and their severity. In order to identify changes in the components of the aforementioned relationship, the evaluation study was designed to monitor police activity on the project roads, estimate changes in road users’ apprehension and behaviour, and assess changes in traffic accidents that might be attributable to the project performance. Thus, the follow-up study of the 700-project consisted of three main parts: Page 190 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

1. monitoring of everyday police operations on the project roads (shift deployments, activity types, citation types and locations, etc.); 2. periodic evaluation of the project’s influence on driver behaviour (through field observations, driver questionnaires and speed measurements); 3. the evaluation of changes in accident numbers and severity within the project area. As the present report focuses on the economic evaluation of the enforcement project, only data on the police activity during the project and changes observed in accident numbers will be further discussed. A description of changes in drivers' behaviour and attitudes which were observed during the project's performance and served as intermediate indicators of the project's effects can be found in Hakkert et al (2001).

3 Target Accident Group

The target accident group of the enforcement project included all severe accidents, i.e. accidents of all types, with fatalities or serious injuries. All accident types were considered as the enforcement was of a general nature, aimed at improving drivers' obedience to most traffic rules. The results pertaining to severe accident counts are considered as the most appropriate to the project's objective. The accident changes were considered on the project's roads (see Table 1). The detailed consideration of the police patrol data (during the project's performance) revealed that shortly after the project's beginning, the actual project territory was reduced in comparison with the original plan, and comprised in fact about 600 kilometres of roads. Table 1 provides the actual lengths of the project road sections, whereas the changes concerned roads No. 70, 77, 75 and “4-north” (these are not the highest priority roads). The accident analysis and the analysis of police activity on road sections addressed these actual lengths of the project roads.

4 Assessment results

4.1 Monitoring of police activity

A special information system was established to monitor the enforcement activity during the project. This included the data of policemen’s shift activity reports of all the NTP subdivisions involved in the project – a monthly input of some 4,500 records. The policeman’s activity report provides details on patrol car locations, activity types and citation categories produced during the shift. Using these data, three groups of summary indices were estimated: (a) inputs - the number of police officers, patrol vehicles and devices per site in a definite time interval (day, week, month); (b) outputs - the level of actual police presence and the citations given; and (c) the efficiency indices, e.g. the performance-against-plan ratios and utilization of resources. The summary indices illustrated the police activity in the form of daily and average monthly figures, with respect to road sections, police regions and the entire project area. The intensity of the police enforcement over the 700-project is characterized by the following facts:

Page 191 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

• the number of patrol units within the project area during a regular weekday shift was about 60; during a similar weekend shift – more than 30; during a night shift – 9.5 and 12, accordingly; • the average monthly amount of the patrol units in all the shifts was some 4,000 in the whole interurban area, of which some 2,700 in the project-area (70%) – Figure 19; • on average, every road section of the 700-project was patrolled about 1,760 hours monthly with a “production rate” of 0.83 citation per shift hour, or 1.2 citations per actual enforcement hour; • the average monthly amount of citations in the project area was more than 24,000 (some 82% of the total), with 1460 citations per project road, on average; • the productivity of a patrol unit in the project area was on average 9 citations per shift, and 7.7 in the whole territory under the NTP responsibility. (The figure does not include automatic citations, produced by F6 – photo radar camera, and Marom – an infra-red speed and gap-following camera).

Figure 19:Monthly number of patrol units in the course of the project (estimated amount of vehicle-shifts in three daily shifts) 5000

4500

4000

3500

3000

2500

Vehicle-shift 2000 All roads under the NTP 1500 responsibility the project area 1000

500 month 0 4/97 5/97 6/97 7/97 8/97 9/97 1/98 2/98 3/98 10/97 11/97 12/97

As became evident from the evaluation of most input/output indices [HAKKERT et al, 1998], project intensity did not stay constant over the whole year. There was an initial period of increasing activity; a fall in September; some priority changes in October and a return to routine in November, however, with a lower intensity in comparison with the initial project period. There were also other factors pointing to two periods in the project’s performance: the intensive publicity accompanied the project only in the first four months; and in September, the central NTP subdivisions were restructured administratively. Finally,

Page 192 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL) a detailed consideration of police deployment (the force split between the project road sections) determined two periods: I - from April to August, II - starting from September 1997 till the project’s completion. This was the first finding taken into consideration in the following accident analysis. The second important point that influenced the accident analysis was the fact that not all road sections were characterized by the same enforcement intensity during the project. Based on three criteria: 1. the number of patrol units per road-km per month (with a threshold of 4); 2. the rate of NTP forces allocated to the road section (5% was accepted as a threshold); 3. the amount of net enforcement hours on the section as opposed to the average value (above average was taken as considerable). Seven road sections, out of fifteen, were chosen as having higher police presence during the project.

4.2 Accident analysis

To assess the NTP project’s influence on safety, the trends in road accidents during the project year were analysed. An evaluation method, which combines both the odds-ratio and a longitudinal (time-series) analysis, was developed. Using this method, the longitudinal models were fitted to the monthly accident counts in the “before” and the “after” periods, for both the treatment and the comparison-group roads, followed by a comparison of the changes. Unlike the classical methods in which the odds-ratio considers the average behaviour “before” and “after”, this method produces odds-ratios for each time point (month) of the “after” period. For the accident analysis, all the roads under the NTP supervision were divided into eight groups, according to following three characteristics: 4. Belonging to the 700–project (yes/no); 5. Police activity level within the project area (high/low presence); 6. Geographical zone (north, centre, south). The third characteristic was added as the geographical regions differ in their traffic patterns and, consequently, in the police activity modes. (Not nine, but eight groups were considered, as in the south there were only two road groups: “non-project roads” and “project roads with higher police presence”.) The roads, which did not belong to the project, served as comparison groups in the corresponding geographical area. The data file for the analysis consisted of the accident records from January 1995 till March 1998. The observed monthly counts of severe accidents, both for the project and comparison group roads, are given in the Appendix. Models for the “before” and “after” periods, were fitted for each road category. A generalized linear model was fitted to the monthly accident counts using the GENMOD procedure of SAS, assuming a Poisson distribution and allowing for over-dispersion. Each model includes a trend and seasonal component. More details can be found in [HAKKERT et al, 2001]. Based on the fitted model for the "before" period for each month of the project period, the expected number of accidents, had there been no intervention, were predicted. Then, a model based on the actual data for the "after" period was also fitted. The monthly odds is Page 193 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL) the ratio of the predicted number of accidents according to the model based on the "after" period and the forecast according to the model based on the "before" period. These monthly odds were evaluated for all the roads (comparison and treatment groups). The next stage of evaluation - the odds ratio - is required to account for possible changes that were not necessarily caused by the intervention (project). Thus, for each month the odds of the treatment group were divided by the corresponding odds of the comparison group to obtain a monthly odds-ratio. This odds ratio was expected to be significantly less than 1, had the project been effective. The "gain" (or loss in accident number due to the treatment) was expressed as the difference between the product of the odds-ratio times the actual count. Table 69 provides the evaluation results. It was seen that since the project started, an increase in accident numbers was observed in most road groups. However, the comparison of the “during the project” accident counts with the “before” period (Table 69, “after/before ratio”) revealed that none of the changes appeared to be significant. A further comparison of the changes observed for the project roads with those occurred in the proper comparison groups, demonstrated that (Table 69, “Odds-ratio”): • A statistically significant reduction of severe accidents, as opposed to the comparison group, was found on the highly enforced road sections in the centre of the country (mainly during the second project period); • No other statistically significant results were obtained. However, as can be seen in Table 69, in most cases the mean value of the odds ratio is much less than one and the average gain (the number of accidents prevented due to the project) is positive. The summary changes in severe accidents over the project's period, in terms of the odds ratio and the "gains" estimated, are highlighted in Table 69. The supplementary analyses performed for all injury accidents and for the numbers of severe casualties (serious injuries and fatalities together) provided similar results [HAKKERT et al, 1998]. To note, separate consideration of fatalities did not bring a significant contribution to the findings, as, due to scarce statistics, the confidence intervals for the odds-ratio values were very wide. None of the project road groups demonstrated a statistically significant change of all injury accidents [HAKKERT et al, 1998]. In general, it was concluded that a general deterioration in safety on the interurban roads occurred during the project year. The phenomenon was less tangible on the project roads, and this result ought to be, due to the concentrated police enforcement applied in this area [HAKKERT et al, 1998].

Page 194 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

Table 69: Odds ratios and estimated "gains" for severe accidents on the project roads Evaluation After/before After/before Odds ratio Estimated Observed period(1) ratio (odds) for ratio (odds) for “Gain” (2) Accident the project comparison- Count (3) roads group roads Road Group: North, Higher Police Presence I1.75 1.29 1.35 -7.74 29 (0.93;3.27) (0.90;1.86) (0.66; 2.79) (-19.09; 15.67) II 1.37 1.41 0.97 1.02 36 (0.75;2.50) (0.99;2.00) (0.49; 1.95) (-18.18; 39.48) Whole 1.50 1.37 1.10 -5.92 65 Year (0.88;2.54) (1.00;1.86) (0.60; 2.02) (-33.88; 45.61) Road Group: Centre, Higher Police Presence I0.95 1.42 0.67 28.58 58 (0.70;1.29) (0.94;2.15) (0.40; 1.12) (-6.07; 86.64) II 0.76 1.32 0.57 58.02 76 (0.57;1.02) (0.89;1.96) (0.35; 0.93) (5.52; 143.62) Whole 0.82 1.36 0.61 88.06 134 Year (0.64;1.06) (0.96;1.92) (0.39; 0.93) (9.85; 208.44) Road Group: South, Higher Police Presence I0.85 1.06 0.80 4.18 16 (0.47;1.53) (0.69;1.64) (0.38; 1.66) (-6.53; 26.50) II 0.95 1.23 0.77 6.48 22 (0.54;1.70) (0.80;1.89) (0.38; 1.59) (-8.24; 36.67) Whole 0.91 1.17 0.78 10.74 38 Year (0.55;1.51) (0.80;1.70) (0.42; 1.46) (-12.24; 53.72) Road Group: North, Lower Police Presence I0.97 1.29 0.75 14.16 41 (0.63;1.48) (0.90;1.86) (0.43; 1.31) (-9.91; 56.26) II 1.25 1.41 0.89 11.58 90 (0.86;1.81) (0.99;2.00) (0.53; 1.47) (-29.21; 79.39) Whole 1.14 1.37 0.83 26.51 131 Year (0.82;1.59) (1.00;1.86) (0.53; 1.31) (-31.67; 118.13) Road Group: Centre, Lower Police Presence I0.79 1.42 0.56 10.56 13 (0.31;2.00) (0.94;2.15) (0.20; 1.54) (-4.66; 52.51) II 0.65 1.32 0.49 19.58 19 (0.27;1.60) (0.89;1.96) (0.19; 1.31) (-4.57; 83.76) Whole 0.70 1.36 0.52 30.38 32 Year (0.32;1.55) (0.96;1.92) (0.22; 1.22) (-5.94; 116.43) (1) The project periods: I (first) April-August 1997; II (second) September 1997-March 1998. (2) “Gain” corresponds to loss in the accident number due to the project. (3) The observed accident counts for the "before" and "after" periods, for the treatment and the comparison group roads are given in the Appendix.

5 Cost-Benefit Analysis

5.1 General

In this section, a Cost-Benefit Analysis (CBA) of the enforcement project is performed. The CBA compares the measure's benefits with the measure's costs, where both values are brought to the same economic framework. Due to the fact that a certain level of enforcement activity was available on the roads prior to the project’s beginning and, therefore, somehow contributed to the safety of the rural road network, the CBA will focus on the changes associated with the project’s performance. In other words, the CBA will

Page 195 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL) compare the additional costs that were invested in the enforcement project's performance with the safety benefits observed. As the time halo-effect of the enforcement project is usually limited, both the costs and the benefits are considered for the project's period only (one year). No conversion to the present economic values is necessary.

5.2 Costs

The additional costs, which were required for the police project's performance, are as follows: 1. personnel costs, including overhead; 2. vehicle fleet expenses; 3. publicity costs. The additional personnel costs were associated with an increase in the police staff, which was needed for the project's performance. A comparison of the numbers of monthly vehicle-shifts during the project with similar data for the "before" period demonstrated that the project's figures were 1.4-1.6 times higher [HAKKERT et al, 1998]; we shall apply an average increase of 1.5 times. Based on the average figure of 1760 hours of patrolling per road per month (see Section 4.1) for the 15 project's roads, the total person-hours during the project month will be 26,400. Applying the norm of 180 person-hours per month, 147 policemen appear to be involved in the project's performance, of whom 49 compose the addition (providing a higher than usual police presence on the project's roads). The personnel costs of one policeman are estimated at 150,000 NIS per year23. A 100% overhead should be added to this figure, accounting for the command, logistics, support staff, equipment’s maintenance, citations' processing, etc. Thus, the additional personnel costs for the project's performance were: 49 policemen * 150,000 * 2 = 14.7 million NIS (at 1997 prices) The vehicle fleet was extended by 10 cars and 3 motorcycles for the project's performance. The cost of a new car is $ 20,000 and a new motorcycle costs $10,000 (as each vehicle stays in use for 5 years, on average, 20% of the initial investment belong to the project’s costs). The annual maintenance expenses of the traffic police in 2003 were 93,000 NIS per a car and 15,000 NIS per a motorcycle. (All the estimates were provided by the Traffic Department of the Police.) Thus, using the average rate $1 = 3.45 NIS (in 1997) and accounting for the change of price index over the years 1997-2002 (by 1.1986), the additional expenses on the vehicle fleet can be estimated as: 10 cars * [0.2 * 20,000 $ * 3.45 + 77,590] = 913,900 NIS on cars, and 3 motorcycles * [0.2 * 10,000 $ * 3.45 + 12,515] = 58,245 NIS on motorcycles (both figures are at 1997 prices). The costs of publicity that accompanied the project were 5.0-7.0 million NIS24 (at 1997 prices).

23 Provided by the Police Traffic Department 24 Provided by the National Road Safety Authority

Page 196 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

Besides, the additional costs of administration were considered to account for the processing of extra citations that were produced during the project. As HAKKERT et al. (1998) found, the productivity of the patrol units increased during the 700-project in comparison with the previous years. The increase in the number of citations would mean extra work for the prosecution of offenders by the police and, in some cases, by courts. However, as known, the fines paid for traffic law violations produce revenues to the treasury. As believed, both figures (of the additional costs and the benefits) are somewhat similar and compensate each other’s effects. The exact figures are unknown and cannot be easily tracked. Therefore, neither costs nor benefits stemming from the extra citations were accounted for in our case.

5.3 Benefits

The project's benefits came from the accidents prevented due to concentrated police enforcement. The value of benefits is estimated as the product of the number of accidents saved and the average accident cost. In the current evaluation the severe injury accidents are considered, as both corresponding to the project's purpose and providing more significant results (see Section 4.2). The number of accidents saved due to the project can be estimated in two ways: 1. Summarizing the values of "gains" estimated by the fitted models (see Table 69). The values are summed up through the project area, i.e. over the five groups of the project's roads (see Table 69). This case will be called "the best estimate". 2. Applying the values of odds-ratio, i.e. the safety effects estimated (see Table 69), the number of accidents prevented is assessed by multiplying the value of the safety effect by the number of accidents observed on project roads during the year "before". The total number of accidents prevented presents a sum of the values from the five groups of the project roads. In this case, a “conservative estimate” of benefits is provided (as less accounting for the general increasing trend, which was observed in the accidents on the whole network of interurban roads during the project year). The details of both estimates are given in Table 70. The "best estimate" states that 150 severe accidents were prevented due to the project's performance. The “conservative estimate” will be 108 severe accidents prevented.

Page 197 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

Table 70: Estimating the number of severe accidents prevented due to the project's performance

Project's road group North, Centre, South, North, Centre, Total higher higher higher lower police lower police police police police presence presence presence presence presence Estimated "gain" ("best -5.92 88.06 10.74 26.51 30.38 149.77 estimate") Estimated safety effect* +10% -39% -22% -17% -48% N/a Observed "before" 62 157 48 140 38 445 accident counts** Number of "saved" -6.2 61.23 10.56 23.8 18.24 107.63 accidents (“conservative estimate”) *Percentage of accident reduction attributed to the measure ** Over the period April 1996-March 1997 In the current Israeli practice, the average accident cost is estimated as a sum of injury costs and damage costs of an average accident in the target accidents’ group. The injury costs are a sum of injury-values multiplied by the average number of injuries with different severity levels, which were observed in the target accidents’ group. The road accident injury values are usually taken as $ 500,000 per fatality, $ 50,000 per serious injury, and $ 5,000 per slight injury [HAKKERT and GITELMAN, 1999]. The damage value is stated as 10% of the injury costs. The above values of injury should be treated as conservative because a recent evaluation of losses from road accidents in Israel recommended a higher estimate of the fatality- value, of $ 930,000 [MATAT, 2004]. The latter accounts for both lost output and human costs, i.e. accounts for the ‘willingness-to-pay’ approach. Table 71 illustrates the calculation of injury costs for an average severe accident observed on rural roads over the year 1997. The injury costs of an average severe accident are NIS 663,815; with the addition of damage-costs, the value of average severe accident is NIS 730,196 (at 1997 prices).

Table 71: Estimating injury costs for an average severe accident on rural roads in 1997

Value Fatality Serious injury Slight injury Total number of injuries in severe 288 1285 1494 accidents The number of severe accidents 1122 1122 1122 Average number of injuries per 0.257 1.145 1.332 accident Injury values, $ 500,000 50,000 5,000 Total injury costs of average $ 192,410 or NIS 663,815 severe accident (at 1997 prices)* *$ 1 = 3.45 NIS

Page 198 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

5.4 Computation of the Cost-Benefit Ratio

Table 72 illustrates the calculation of the cost-benefit ratio (CBR) of the enforcement project. The total value of the project's costs was of 21-23 million NIS (at 1997 prices), or about 6 million Euros (at 2002 prices). The total value of the project's benefits was of 79- 109 million NIS (at 1997 prices), or of 21-29 million Euros (at 2002 prices). Consequently, the value of CBR was better than 1:3.5 for the "conservative estimate" of the accidents prevented and about 1:5 for the "best estimate". For the range of cost and benefit assumptions considered, the enforcement project appears to be cost-effective. Table 72: Costs and benefits of the enforcement project considered

Costs Benefits "Best "Conservative estimate" estimate" Personnel, with overhead, 14.7 Number of severe 150 108 million NIS accidents saved Vehicle fleet, million NIS 0.914 + 0.058 Average accident 730,196 730,196 cost, NIS Publicity, million NIS From 5.0 to 7.0 Total, million NIS (1997) From 20.672 to Total, million NIS 109.53 78.86 22.672 (1997) Total, million Euro (2002)* From 5.53 to Total, million Euro 29.3 21.1 6.07 (2002)* Cost-benefit ratio From 1 : 5.3 From 1 : 3.8 to 1 : 4.8 to 1 : 3.5 *Change of price index over 1997-2002 is 1.1986. In 2002: 1 Euro = 4.48 NIS.

6 Decision-Making Process

The follow-up study of the enforcement project was initiated by the National Road Safety Authority. The study's steering committee included representatives from the Ministry of Interior Security, National Road Safety Authority and the Police Traffic Department. The evaluation results were reported to the Head and other high level decision-makers of the Road Safety Authority and to the Traffic Police Command. The follow-up consideration of the enforcement project concerned mostly the changes in actual driver behaviour, drivers' attitudes and accident numbers. As the majority of accident changes observed on the project's roads were statistically not significant, the project's results were stated as "moderate" [HAKKERT et al, 2001]. Such a "conservative" estimate was given to the project also accounting for a decrease in the project's intensity during the second half of the project's year, changes in the force deployment over the project's year, and lack of a strict policy in the enforcement modes applied by the different police units. One of the main reasons for the limited success was seen in the gap between the project target and everyday enforcement activity. The recommendations were given to develop more focused enforcement operations, i.e. shorter in time, more concentrated in area/enforcement subject, and more flexible in performance by the police units [HAKKERT et al, 1998]. To note, in the coming years, 1998-1999, a series of short-term enforcement experiments was performed by the Israeli Traffic Police.

Page 199 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

The cost-benefit analysis presented in this report actually rehabilitates the 700-project demonstrating that in spite of the doubts as to the significance and consistency of the results attained, the project was definitely beneficial from the economic viewpoint. It is believed that a repeat discussion on the project's results with the decision-makers, and especially with the Police Command, will stimulate the performance of other projects of intensive police enforcement.

7 Role of barriers

None of the known barriers to the use of the efficiency assessment tools [WP2, 2004] played a serious role in the CBA of the enforcement project considered. Both the Traffic Department of the Police and the National Road Safety Authority assisted in collecting the data to perform the economic evaluation. The police enforcement project was initiated by the Israeli authorities based on the international experience that proved the effectiveness of such a measure for improving drivers' behaviour and road safety. Therefore, neither institutional nor implementation barriers to the EAT application seem to be relevant in this case. The technical barriers, e.g. lack of knowledge of safety effect, were overcome by means of relevant data collection and fitting statistical models for various evaluation needs.

8 Discussion

The concentrated general police enforcement project took place for a whole year on the most heavily travelled interurban roads in Israel. The project aimed at a reduction in severe accidents on the roads in focus and, concurrently, at an improvement in the Traffic Police working modes. The project did not attain its full purpose, as a significant reduction of severe accidents was found only on one of the five project road groups. However, in four of the five project road groups the mean value of the odds ratio was much less than one, indicating a positive average safety effect. The economic evaluation based on the average values of safety effects demonstrated that the enforcement project was beneficial. An important finding of this study is that had the cost-benefit analysis been performed immediately after the police project completion, the conclusions of the evaluation study would have been more optimistic than those given in the report by Hakkert et al. (1998). The CBA compared the additional costs, which were required for the police project performance with the safety benefits (severe accident savings) attained. The CBA presented in this study can be characterized as follows: • the evaluation findings support the measure's implementation; • to estimate the safety effects statistical models were fitted to the accident data and the evaluation was in line with the criteria of correct safety evaluation [WP3, 2004]; • the accident costs were fitted to the accident type considered, however, they should be treated as conservative as the injury costs did not account for the ‘willingness-to-pay’ component; • the measure does not have a long-term effect, therefore both costs and benefits were considered for the year of implementation only;

Page 200 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

• the evaluation study was initiated by the authorities and the results were accepted by the decision-makers; • the barriers for the CBA's performance did not play an essential role in the case presented. The limitations of the CBA performed are as follows: • the calculation of benefits was based on mean values of safety effects, whereas part of them were not stated as statistically significant; • the economic analysis considered the benefits stemming from the project's safety effect only. Neither environmental impact nor mobility effect was quantified, as the influence of the enforcement project appears to be insignificant in this sense.

Page 201 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

References

BJORNSKAU, T. and ELVIK, R. (1992): Can road traffic law enforcement permanently reduce the number of accidents? Accident Analysis & Prevention 24, 507-520. CAMERON, M., NEWSTEAD, S. and GANTZER, S. (1996) Effects of enforcement and supporting publicity programs in Victoria, Australia. Proceeding of International Conference on Traffic Safety on Two Continents, Prague, Czech Republic; VTI konferens 4A, part 4, pp. 244-253. FITZPATRICK, K (1992): A Review of Automated Enforcement. Compendium of Technical Papers, Institute of Transportation Engineers, pp.184-188. HAKKERT, A. S., YELINEK, A. and EFRAT, E. (1991): Police surveillance methods and police resource allocation models. In Enforcement and Rewarding: Strategies and Effects, eds M. J. Koornstra and J. Christensen, pp. 98-101. SWOV, Leidschendam, the Netherlands. HAKKERT, A.S., GITELMAN V., COHEN, A., DOVEH, E., UMANSKY, T., SHINAR, D. (1998): A Follow-up Study of a New Deployment of the National Traffic Police in 1997 - Focused Police Enforcement. Research Report No. 98-268, Transportation Research Institute, Technion, Israel (in Hebrew). HAKKERT A. S. and GITELMAN V. (1999): Development of a National Road Safety Program in Israel: Baseline, Components and Lessons. Proceedings of Int. Conf. Traffic Safety on Two Continents, Malmo, Sweden; VTI konferens 13A, part 3, pp. 75-92. HAKKERT, A.S., GITELMAN V., COHEN, A., DOVEH, E., UMANSKY, T. (2001): The evaluation of effects on driver behaviour and accidents of concentrated general enforcement on interurban roads in Israel. Accident Analysis and Prevention 33, pp. 43- 63. MATAT (2004): Road Accidents in Israel: the scope, the characteristics and the estimate of losses to the National Economy. MATAT - Transportation Planning Centre ltd, Ministry of Transport. OECD (1999): Enforcement. Chapter 5 in Safety Strategies for Rural Roads. Road Transport and Intermodal Research, Organisation for Economic Co-operation and Development, IRRD No 491006, Paris. OEI, H.L. (1998): The Effect of Enforcement on Speed Behaviour; A Literature Study. Proceedings of International Conference ‘Road Safety in Europe’, Bergisch Gladbach, Germany; VTI konferens 10A, part 10, pp.107-118. ZAAL, D. (1994):Traffic Law Enforcement: A Review of the Literature. Report No.53, Accident Research Centre, Monash University, Australia. ZAIDEL, D. M., HOCHERMAN, I. and HAKKERT, A. S. (1994): Evaluation of a National Traffic Police Force, Transportation Research Record 1401, Transportation Research Board, Washington, D. C., pp.37-42. WP3 (2004): Improvements in efficiency assessment tools. ROSEBUD. WP2 (2004): Barriers to the use of efficiency assessment tools in road safety policy. ROSEBUD.

Page 202 INTENSIFICATION OF POLICE ENFORCEMENT (SPEED AND ALCOHOL)

Appendix: The observed counts of severe accidents for the road groups considered Non-project road groups Project road groups Year Month North Centre South North, Centre, North, Centre, South, lower lower higher higher higher police police police police police presence presence presence presence presence 9513319111746115 2 18 9 15 12 5 11 9 3 3 22 9 13 15 5 9 12 4 4 25 7 14 13 3 8 11 4 52312221258103 6 32 9 15 20 3 8 13 7 7 26 11 22 16 3 12 13 9 8 31 8 12 16 2 1 11 5 9251310130383 1025121995584 11251117725134 12 30 14 13 16 3 3 15 5 9612810141048137 22991694972 3201291514117 418417152793 52217151104145 6 27 8 21 11 4 6 10 9 7271214854134 829721764163 9111171555135 10 22 12 10 13 4 6 16 4 11 23 6 13 10 7 7 12 3 12 21 9 4 14 0 6 16 4 97118811122692 2 15 5 19 14 3 4 17 1 3 19 6 10 10 0 3 12 5 422912111392 52114141122184 630819669103 73413181049125 8 26 7 17 14 1 9 18 4 9 27 6 19 10 3 4 13 3 101151570394 11 24 7 13 18 1 1 10 3 12 36 10 10 13 5 9 14 4 98 1 29 9 16 14 3 7 11 4 21718161217112 323137166582

Page 203 CASE J1: 2 + 1 roads in Finland

ROSEBUD WP4 – CASE J REPORT

2 + 1 ROADS IN FINLAND

BY MARKO NOKKALA,

VTT BUILDING AND TRANSPORT, FINLAND 2 +1 ROADS IN FINLAND

TABLE OF CONTENTS

1. PROBLEM TO SOLVE ...... 207 2. DESCRIPTION OF THE MEASURE...... 207 3. TARGET GROUP ...... 208 4. ASSESSMENT METHOD...... 209 5. METHOD OF ANALYSIS...... 209 6. ASSESSMENT QUANTIFICATION...... 210 7. ROLE OF BARRIERS ...... 211 8. DISCUSSION...... 211

Page 205 2 +1 ROADS IN FINLAND

CASE OVERVIEW

Measure The measure is to construct 2+1 roads with pavement in the middle of a narrow highway. Problem Head-on collisions have been frequent in traffic in Finland and Sweden, with the number of fatalities resulting from the accidents increasing as a proportion of total road accidents. Target Group All the road users driving on Finnish roads and highways where new 2+1 construction takes place. Targets The safety measures applied in the middle of the road have two main objectives: To avoid the head-on collisions of the vehicles off their lane and to reduce the accident severity of the remaining crashes. Initiator In Finland, the national road authorities have been responsible of road construction to produce the required alterations and investments for 2+1 roads. The discussion on the safety impact of the measure has been centred on by a few key experts, otherwise the measure is not widely promoted. Decision-makers Decision-makers are usually located in the headquarters of the national road authorities, but if 2+1 roads are part of major national level investments, then the participation of national level (i.e. Ministry) decision-makers is required. Costs For Finland, the total costs of the 2+1 road construction have so far been 417 million Euros, which has resulted in over 500 kilometres of 2+1 road constructions, mostly as short overtaking sections rather than a full road length. Benefits The main benefit from implementing the measure consists of an important reduction of the number of head-on collisions at a cost lower than that of highway construction. These benefit mainly national level decision-makers, insurance companies and those road users who still experience a traffic accident, but with less damage than in the absence of the 2+1 road. Cost-Benefit Ratio The Cost-Benefit Ratio is 1.25.

Page 206 2 +1 ROADS IN FINLAND

1 Problem

Both Finland and Sweden have committed to a version of zero-tolerance for traffic accidents with fatalities. Whilst it has been realised that the random variable in causing the accidents cannot be controlled, several studies have addressed the efficiency of various measures in preventing accidents (see, for instance, Peltola and Wuolijoki (2003) and Kärki et al. (2001) for reviews of such typologies). However, these studies do not list the construction of 2+1 roads into road safety measures list, which has only been a recent addition (see Annex for latest available cost-effectiveness information on road safety measures in Finland). Head-on collisions have been one of the major causes of fatal accidents on roads in Nordic countries. In 1998 the Swedish National Road Administration, Vägverket, started a development project on 13-meter wide roads. The rationale for the program was that the construction of a wider road would be a cost-effective way of increasing road safety on highways, compared to alternative measures resulting in the same net effect in prevention of crashes and fatalities. Finland has been slow to use the 2+1 road as a solution to prevent the accidents. The National Road Administration in Finland has studied the Swedish case with close interest, but the frequency of adopting the measure has not been transferred to Finnish context. In Sweden, however, there have been frequent campaigns to show the cost-efficiency of this measure [NTF 2003]. It has been estimated that one kilometre of 2+1 road costs one- twentieth of the cost of one kilometre of motorway, which has been translated into an argument that building a motorway is not in fact an efficient way to increase road safety compared to the 2+1 road solution. Head-on collisions have been a severe problem as a percentage of total fatalities in Finland; between 1996-2000 an average of 80-85 percent of fatal accidents on two-lane highways were due to head-on crashes. For Sweden, figures for 1993-2000 show that 140 people were killed in head-on collisions, while the number of severely injured was 450 for the same period. For this case study a comparative study of Finnish and Swedish experiences were chosen to study the decisions made in choosing the 2+1 road as means to prevent accidents. Particularly in Finland, the decision-making on road construction does not take into consideration a pure road safety aspect; the decisions are always based on the socio- economic profitability of the project, where traffic safety is only one of the dimensions. This is explained in greater detail in the section dealing with the assessment method.

2 Description of the measure

2+1 road construction is a measure where an existing road is updated to have a middle lane changing direction every 1-2.5 kilometres. Of course, alternatively the construction method can be applied to new road sections as well, but since the upgrading is a low-cost measure compared to, for instance, construction of a new motorway, the standard application is to existing road sections. In principle, the 2+1 road construction takes place on 13-meter-wide roads, and it is considered as means of upgrading other solutions, mainly wide shoulders or wide lanes. Figure 20 illustrates the principle difference between these three approaches. Between the three different approaches, the distinctive advantage of the 2+1 solution is that it prevents head-on collisions, whereas wide shoulders and wide lanes allow for greater driving

Page 207 2 +1 ROADS IN FINLAND margins and can prevent crashes out of the roads with more margins. As noted, the most effective crash reduction will result from the reduction in head-on collisions. Figure 20: Construction possibilities for 13-meter wide road: Wide shoulders, wide lanes and 2+1 design

Source: Larsson et al. The road construction must also pay great deal of attention to switching the overtaking lane from one side to another. The principles of this are shown in Figure 21. There are signals both on the road and on the side to indicate the width of the overtaking lane to the other side. Figure 21: Designing the lane transition zone

Source: Larsson et al.

We will limit the analysis in this case study to the case where the 2+1 road is constructed with fixed median cable, which is also the most common way of constructing the 2+1 roads. The more recent promotion of 2+2 road is obviously even more effective way of increasing road safety in Nordic roads, but there is limited data from both countries in terms of the impacts of these roads.

3 Target group

The construction of new roads using the 2+1 design is likely to benefit all road users, with the reduction of head-on collisions. For those road users who cause the possibility of an Page 208 2 +1 ROADS IN FINLAND accident, for instance by sleeping behind the wheel, the immediate benefit is the reduction to zero in the probability to have a serious head-on collision. The measure is considered effective in fully preventing the head-on collision. For the drivers who are in danger of facing a possible head-on collision, the benefit is derived from the fact that the probability again is diminished by the existence of a middle road cable.

4 Assessment method

Sweden has been ahead of Finland in designing and implementing the 2+1 road construction. There is a relatively well-documented, existing database on the constructed sections of the 2+1 roads produced by VTI, the Swedish National Institute for Transport Research. In Finland, the choice of methods for upgrading roads, particularly from the safety point of view, has been unsatisfactory, focusing on speed improvements over low- cost safety measures. Finnish aggregate data on accidents shows that the cost-effectiveness of the 2+1 road is of average in terms of cost for accident prevention [NOKKALA and PELTOLA 2004]. So far, 575 kilometres of road with the 2+1 structure has been built at the cost of 417.6 million euros. The associated reduction in fatal accidents is estimated to be around 29 accidents annually, and in accidents resulting in death an average of 5.5 accidents are prevented annually. In discussions with the representatives of National Road Administration, a case study for the CBA was selected to represent a typical project focusing on the 2+1 road construction. For Finland, the TARVA program can be used for assessing the accidents on road networks. TARVA contains detailed information on the road network based on road addresses and information on investment projects (broken down by components) and accidents data assigned to road address. This model can be used to analyse the reduction in accidents for each given section of the road network.

5 Method of analysis

Finland uses a specific procedure for evaluating the transport projects: socio-economic profitability analysis. This is a method that combines both quantitative and qualitative techniques, but is very much based on CBA. The following components of the calculation need to be produced, and the Ministry of Transport and Telecommunications has published a set of official values to be used: • Accident costs • Time savings • Vehicle costs • Emissions • Noise • Maintenance costs • Investments The standard methodology can be applied to the case of 2+1 road, but a word of caution is needed. The method tends to heavily stress the role of time saving component, often overlooking other dimensions of analyses. Relying on cost-effectiveness methods could be more appropriate, but it is against the evaluation principles. Therefore, we apply the

Page 209 2 +1 ROADS IN FINLAND standard methodology keeping in mind the constraints. Table 73 below shows the unit values used in analyses. Table 73: Unit values for various components on the socio-economic profitability analysis COMPONENT UNIT PRICE VEHICLE COST LIGHT/HEAVY 24.7/84.8 VEHICLES PER KILOMETRE (+VAT), CENTS Time savings light/heavy vehicles per 10.6/26.7 km, cents Severe accident € 386,832 Accident, death € 2,430,316 Accident, average € 84,094 Emission costs per ton, average *SO2 8,322 *NOX 734 *PM2,5 103,537 *CO 16 Noise (annualised cost per inhabitant) 959 € Source: Finnish Ministry for Transport and Telecommunications Figures in Table 73 are officially used in all Finnish transport project appraisals and their unit values are confirmed by the Ministry of Transport and Telecommunications.

6 Assessment quantification

The calculations have been complicated by the fact that the construction of road sections in Finland is taking place on the terms of developing the road as a whole, as opposed to constructing a separate measure, such as the pavement alone. In fact, in many cases the centre of the road pavement is part of an upgrading of existing road, where traffic volumes have increased to degrade the existing road. On the other hand, as explained in the previous section, the use of the socio-economic profitability calculus allows one to take into consideration the full impact of the investment, including the changes in speed. For the calculation we have applied the socio-economic profitability analyses, with a 5 percent discount rate and maintenance period of 20 years. The technical durability of the 2+1 road is most likely less than 20 years, as new methods to construct motorways with lower costs create pressures to upgrade the roads eventually. The Finnish case is from Highway Nr. 4 (VT 4), from the section between Lahti and Heinola. This particular section of the road is considered one of the rare congested highways outside the Helsinki metropolitan area. The road was constructed into 2+1 format in 1993 and the section is 26 kilometres long (Hiltunen 2004). However, the road is constructed without the median cable, as the solution is from 1993 when the construction was not fashionable. In the Finnish context the road has a large traffic flow, 12,000 vehicles per day (Tuovinen et al. 2004). Data on traffic volumes and driving times was available from both the pre-investment period and the period during which the road has been operational. Accident risk on the Lahti-Heinola road was estimated for the period 1998-2002 as 1.4 deaths per 100 million driven kilometres and 5.5 million severe accidents per 100 million

Page 210 2 +1 ROADS IN FINLAND driven kilometres. These figures were extrapolated to the period of 1993-1998 and contrasted with data from 1988-1993 to estimate the change in accidents. The safety impact of the measure is considered to be 100% elimination of head-on collisions and the fatalities resulting from these accidents if a median cable is inserted. The median cable is able to fully prevent the accidents with vehicles from other lanes, but can only partially reduce other damages from accidents where vehicles collide with the cable. In monetary terms, however, the size of an average accident is reduced significantly. However, in Finland there exists only one section of the road that has 2+1 construction with a median cable. We were forced to use an example without the median cable, which results in lower accident prevention rate. The total cost of the project was estimated at 11.5 million Euros, which consisted of both upgrading the road and the necessary expansion of the road width. Using the data available, the socio-economic profitability analysis was carried out. The resulting benefit-cost ratio was 1.25. The project was considered acceptable by pure financial terms, even if the benefit-cost ratio was modest. The calculations have not been subject to significant sensitivity analysis, but it can be noted that the main factor that may change the calculus significantly is the change in driving speed, but this in fact is well documented and should not be subject to too much variability.

7 Role of barriers

Barriers to constructing this type of road exist, and in decision-making this appears at all hierarchy levels. Planners avoid the 2+1 solution in the first place because they know that it will face resistance at high levels of decision-making. This is why there has been only one example of the measure so far. New, smaller examples of simply constructing overtaking lanes have adopted the principle of always having the median cable, which is a clear indicator of the observed effect of the cable in preventing collision accidents. Unlike in Sweden, Finnish decision-making seems to take the alternative, but more costly, route of upgrading roads to broader motorways instead of 2+1 roads. This is perhaps in the long-term interest of the government, but it overlooks the important cost factor. Data availability for conducting the case study was good and authorities were helpful in compiling required data. This suggests that the real problem of adopting the method lies outside the authorities and is within the decision-making system.

8 Discussion

The findings show that the results are promising in terms of the expected reduction in head-on-collisions. The value for the Finnish case in the CBA is unexpectedly low, perhaps an indication of the relatively little impact of the safety in the socio-economic profitability analysis. In the case of major investment programs, effects other than safety tend to dominate the analyses and the isolation of the safety impact alone become meaningless, as the project would not have been implemented by constructing the safety measure alone. This is because the construction of 2+1 road requires updating the existing road and possibly carrying out a number of supporting measures to be able to install the median cable. The section between Lahti and Heinola is already in an upgrading process. The 2+1 road will be replaced by a motorway later in 2005. Therefore, obtaining long-term series data on

Page 211 2 +1 ROADS IN FINLAND the safety effect of the 2+1 road will not be possible. This also makes it difficult to interpret the CBA results, where the estimated maintenance time was 20 years, but it now appears to remain around 12 years.

REFERENCES

HILTUNEN, L. (2004): Uusien tietyyppien liikenneturvallisuus. Seminar paper at Helsinki Technical University. KÄRKI, O., H. PELTOLA JA A. WUOLIJOKI (2001): Tienpidon toimien turvallisuusvaikutukset. Tie- ja liikenneolojen hallintajärjestelmän (TILSU) sisältämien toimien arviointi. Tiehallinnon sisäisiä julkaisuja 47/2001. Helsinki. LARSSON, M., T. BERGH JA A. CARLSSON (2003): Swedish Vision Zero Experience. MINISTRY OF TRANSPORT AND TELECOMMUNICATIONS (2003): Guidelines for project appraisal. In Finnish, with English abstract. NOKKALA, M. JA H. PELTOLA (2004): Tienpidon uus- ja laajennusinvestointien kustannustehokkuus liikenneturvallisuuden näkökulmasta (LIIKUTUS). Publication in LINTU-program, the Finnish program for traffic safety. SUMMALA, H. (2003): Kohtaamisonnettomuudet: Pääteiden suurin turvallisuusongelma. In Tiennäyttäjä 6/2003. TARVA (2003). TARVA 4.4 Käyttöohje. Liite 2: Keskimääräiset onnettomuusasteet ja K- arvot. Tuovinen, P., T. Luttinen, Å. Enberg (2004): Traffic Flow Characteristics on main road 4 between Lahti and Heinola in Finland. In Finnish, with English abstract.

Page 212 2 +1 ROADS IN FINLAND

Annex - Measures to improve traffic safety in Finland Toimenp KVL Hinta Hvjonn. Kuolem.Hinta M€/vaikutus- 1.vuod matka aj/vrk yht. vähen. vähen.aikana säästetty tuotto Nro Toimenpide km 1000 € vuosit. vuosit. hvjo kuoll % inv. 921 Kameravalvonta (50%) 597 7461 1889 12,51 2,864 0,01 0,0 502,1 684 Nopeusrajoitus 100 -> 80 km/h 29 2159 13 0,31 0,083 0,00 0,0 1979,6 685 Nopeusrajoitus 80 -> 60 km/h 11 6066 10 0,44 0,058 0,00 0,0 2515,7 676 Nopeusrajoitus 50 -> 40 km/h 2 3885 2 0,10 0,016 0,00 0,0 3171,4 678 Nopeusrajoitus 60 -> 50 km/h 2 2335 1 0,02 0,003 0,00 0,0 1147,9 502 Jäykät pylväät myötääviksi 10 15914 70 0,23 0,055 0,02 0,1 256,9 383 Liikennetieto-ohjaus, valmiit valot 50 5223 162 0,28 0,065 0,04 0,2 131,1 924 Ajosuuntien erottaminen rakent. 297 9013 26653 9,71 2,965 0,14 0,4 33,0 639 Kaiteiden kunnostus 3 7441 59 0,01 0,006 0,21 0,5 27,2 361 Uusi tievalaistus jäykin pylväin 4 12573 224 0,17 0,022 0,09 0,7 43,1 362 Uusi tievalaistus myötäävin pylväin 615 5271 33259 13,98 2,933 0,16 0,8 30,3 601 Koroke päätien suojatielle 0 5479 15 0,01 0,001 0,08 0,8 31,8 631 Kaiteiden rakentaminen 121 6183 6137 2,04 0,359 0,15 0,9 21,8 503 Kallioleikkausten leventäminen 15 5951 782 0,11 0,044 0,35 0,9 15,4 521 Muuttuva nopeusrajoitus 620 16011 24291 10,16 1,563 0,16 1,0 25,6 132 Kevytliikenteen ylikulku 0 5479 57 0,02 0,003 0,15 1,0 20,7 504 Esteiden poistaminen 84 12139 5162 1,89 0,212 0,14 1,2 19,5 638 Liittymämerkintöjen tehostaminen 0 2055 6 0,01 0,001 0,17 1,2 69,0 905 Kapea 4-kaistatie 557 8500 336214 63,46 13,252 0,26 1,3 13,6 342 Linja-autopysäkki maaseudulla 2 3303 62 0,02 0,002 0,17 1,6 15,4 913 Yksityistiejärj. 1185 5368 74331 16,66 2,199 0,22 1,7 12,8 501 Luiskien loiventaminen 214 4912 16677 1,28 0,481 0,65 1,7 8,0 289 Väistötilan rakentaminen 83 3812 9092 1,90 0,265 0,24 1,7 12,2 912 Kevytliikenne rinnakkaisväyl. 159 5780 6553 0,67 0,156 0,49 2,1 7,8 658 Taajaman saneeraus 11 4718 2329 0,66 0,054 0,18 2,2 13,4 922 Mol -> MO 72 12477 163602 4,18 3,329 1,96 2,5 4,7 602 Suojatien valo-ohjaus 0 5479 43 0,02 0,001 0,18 2,9 16,2 261 Lisäkaistan rakentaminen 50 25043 5524 1,74 0,090 0,16 3,1 13,1 634 Reunapaalut, 100 km/h 109 903 368 0,16 0,024 0,45 3,1 26,6 381 Uusi valo-ohjaus, 4-haaraliittymä 4 13387 3798 0,70 0,074 0,36 3,4 9,5 902 Ohituskaistatie+kaide 575 5230 417642 28,95 5,595 0,72 3,7 4,8 288 Kiertoliittymän rakentaminen 9 6531 15521 1,61 0,198 0,48 3,9 5,7 914 Riista-aita, mol 449 7680 14219 2,53 0,166 0,28 4,3 7,9 301 Kiihdytyskaista eritasoliittymään 10 13980 4575 0,32 0,052 0,71 4,4 4,4 282 Liittymän porrastaminen 87 4774 94172 5,88 1,022 0,80 4,6 4,1 281 Keskisaarekkeen rakentaminen 2 5365 419 0,04 0,004 0,50 5,2 5,0 911 Kevyen liikenteen väylän rak. 544 5161 82065 3,54 0,777 1,16 5,3 3,2 285 Nelihaaraliittymän kanavoinnin täydent. 2 5753 847 0,07 0,007 0,65 6,0 4,0 173 Kapean tien leventäminen, maaseutu 1926 2278 284883 15,53 2,177 0,92 6,5 3,2 290 Sivuteiden saarekkeen rakentaminen 4 3484 545 0,05 0,004 0,58 6,8 4,1 283 Liittymän siirto parempaan paikkaan 15 5288 6937 0,39 0,049 0,88 7,1 3,2 133 Henkilöauto & kevytliikenne alikulku 33 6404 69925 2,87 0,468 1,22 7,5 2,6 482 Riista-aita muilla teillä 261 5311 6008 1,05 0,036 0,29 8,3 6,7 382 Uusi valo-ohjaus, 3-haaraliittymä 4 7671 3232 0,21 0,025 1,05 8,6 3,5 172 Suuntauksen parantaminen, maaseutu 618 4325 299068 15,07 1,709 0,99 8,7 2,7 284 Nelihaaraliittymän täyskanavointi 26 4988 26771 1,07 0,140 1,25 9,6 2,3 131 Kevytliikenteen alikulku 72 5588 73054 1,59 0,287 2,30 12,7 1,4 302 Eritasoliittymän täydentäminen 31 17653 46299 3,64 0,168 0,64 13,8 3,2 102 Kevytliikenteen väylän parantaminen 2 3005 298 0,01 0,001 1,35 14,9 1,8 632 Näkemäraivaus 104 3462 467 0,16 0,009 1,00 17,3 14,3 915 Eritasoliittymän rakent. 80 8387 983502 20,62 2,385 2,39 20,6 1,1 286 Kolmihaaraliittymän kanavointi 67 4820 73795 1,00 0,139 3,70 26,5 0,8 287 Liittymän kevyt parantaminen 20 5747 2945 0,35 0,030 2,77 32,7 5,8 923 Yksittäisen ohituskaistan rakent. 138 4219 42294 0,15 0,000 14,58 - 0,1 901 Ohituskaistatie 4 5479 2131 0,12 -0,011 0,86 - 0,8 690 Nopeusrajoitus Kesä 80->100 km/h 5 5619 4 -0,10 -0,032 - - -2359,4 681 Nopeusrajoitus 70 -> 80 km/h 4 7581 1 -0,34 -0,106 - - -31224,0 679 Nopeusrajoitus 60 -> 70 km/h 4 7581 2 -0,34 -0,110 - - -15999,0 903 Leveäkaistatie 65 6324 42866 2,24 -0,153 0,96 - 1,0 683 Nopeusrajoitus 80 -> 100 km/h 69 6924 134 -3,60 -1,051 - - -2364,9 YHTEENSÄ 10240 6091 3297108 246,75 45,133 0,68 3,7 5,0 Source: Nokkala and Peltola, 2004

Page 213 CASE J2: 2 + 1 roads in Sweden

ROSEBUD WP4 – CASE J REPORT

2 + 1 ROADS IN SWEDEN

BY MARKO NOKKALA,

VTT BUILDING AND TRANSPORT, FINLAND 2 +1 ROADS IN SWEDEN

TABLE OF CONTENTS

1 PROBLEM TO SOLVE ...... 217 2 DESCRIPTION OF THE MEASURE...... 217 3 TARGET GROUP ...... 219 4 ASSESSMENT METHOD...... 219 5 METHOD OF ANALYSIS...... 219 6 ASSESSMENT RESULTS...... 220 7 DECISION MAKING PROCESS...... 220 8 ROLE OF BARRIERS ...... 221 9 DISCUSSION...... 221

Page 215 2 +1 ROADS IN SWEDEN

CASE OVERVIEW

Measure The measure is to construct 2+1 roads with pavement in the middle of a narrow highway. Problem Head-on collisions have been frequent in traffic in Sweden, with the number of fatalities resulting from the accidents increasing as a proportion of total road accidents. Target Group All the road users driving Swedish roads and highways, also the drivers who drive the opposite direction (due to the safety effect). Targets The safety measures applied in the middle of the road have two main objectives: to avoid the head-on collisions of the vehicles off their lane and to reduce the accident severity of the remaining crashes. Initiator In Sweden, the national road authorities have been responsible of road construction to produce the required alterations and investments for 2+1 roads. There has been active public discussion on the safety impact of the 2+1 construction, particularly when it has been contrasted with construction of motorways, which are 20 times more expensive per kilometre as opposed to 2+1 road construction. Decision-makers Decision-makers are usually located in the headquarters of the national road authorities, but if 2+1 roads are part of major national level investments, then the participation of national level (i.e. Ministry) decision-makers is required. Costs The average cost per kilometre of 2+1 road in Sweden is 125,000 Euros. Benefits The main benefit from implementing the measure consists of an important reduction of the number of head-on collisions at a cost lower than that of highway construction. This benefits mainly national level decision-makers, insurance companies and those road users who still experience a traffic accident but with less damage than in the absence of the 2+1 road. On the average, every 40 kilometres of 2+1 road construction reduce the probability of fatal accident by one death person. Cost/Benefit-Ratio: The Cost-Benefit Ratio is 2.26 in the Swedish case.

Page 216 2 +1 ROADS IN SWEDEN

1 Problem

Both Finland and Sweden have committed to a version of zero-tolerance for traffic accidents with fatalities. Whilst it has been realised that the random variable in causing the accidents cannot be controlled for, several studies have addressed the efficiency of various measures in preventing accidents (see, for instance, Peltola and Wuolijoki (2003) and Kärki et al. (2001) for reviews of such typologies in Finland). However, these studies do not list the construction of 2+1 roads into the road safety measures list, which has only been a recent addition. Head-on collisions have been one of the major causes of fatal accidents on roads in Nordic countries. In 1998 the Swedish National Road Administration, Vägverket, started a development project on 13-meter wide roads. The rationale for the program was that the construction of a wider road would be a cost-effective way of increasing road safety on highways, compared to alternative measures resulting in the same net effect in prevention of crashes and fatalities. Finland has been slow to use the 2+1 road as a solution to prevent the accidents. The National Road Administration in Finland has studied the Swedish case with close interest, but the frequency of adopting the measure has not been transferred to Finnish context. In Sweden, however, there have been frequent campaigns to show the cost-efficiency of this measure [NTF 2003]. It has been estimated that one kilometre of 2+1 road costs one- twentieth of the cost of one kilometre of motorway, which has been translated into argument that building a motorway is not in fact an efficient way to increase road safety, compared to the 2+1 road solution. Head-on collisions have been a severe problem as a percentage of total fatalities in Finland, between 1996-2000 an average of 80-85 percent of fatal accidents on two-lane highways were due to head-on crashes. For Sweden, figures for 1993-2000 show that 140 people were killed in head-on collisions, while the number of severely injured was 450 for the same period. For this case study a comparative study of Finnish and Swedish experience was chosen to study the decisions made in choosing the 2+1 road as means to prevent accidents. Particularly in Finland the decision-making on road construction does not take into consideration a pure road safety aspect; the decisions are always based on socio- economic profitability of the project where traffic safety is only one of the dimensions.

2 Description of the measure

In principle, the 2+1 road construction takes place on 13-meter wide roads, and it is considered as means of upgrading other solutions, mainly wide shoulders or wide lanes. Picture 2 illustrates the practical application of 2+1 road construction in Sweden.

Page 217 2 +1 ROADS IN SWEDEN

Picture 2: An example of 2+1 road with a median cable

Source: Larsson et al, 2003 As shown in the Picture 2, the most common way to construct the 2+1 road is to set the fixed steel median cable on the road, which then shifts to the other side when the overtaking lane is switched to the other direction. The main problem with the solution is the inability of the road to adjust to changes in traffic flows, for instance during the congestion period as the solution is fixed and cannot be adjusted. Figure 22 shows the other possibilities to utilise the 13-meter width of the road. Wide shoulders mean that the standard lanes are left somewhat narrower, but shoulders have been extended so that driving off the road becomes more difficult. In the case of wide lanes, small errors in steering do not lead to driving off the road, but the shoulders are narrower. In the case of the 2+1 road, shoulders are narrow and the lane width is similar to that of wide shoulder lanes. The overtaking lane in the middle is slightly narrower than the standard lanes. As can be seen, the 2+1 road most effectively reduces head-on collisions, compared to the other two solutions. It is also the most effective solution to deal with congestion, as it allows for overtaking more easily than the other two construction possibilities. Figure 22: Construction possibilities for 13-meter wide road: Wide shoulders, wide lanes and 2+1 design

Source: Larsson et al, 2003 We will limit the analysis in this case study to the case where the 2+1 road is constructed with a fixed median cable, which is also the most common way of constructing the 2+1 roads. The more recent promotion of 2+2 road is obviously even more effective way of

Page 218 2 +1 ROADS IN SWEDEN increasing road safety in Nordic roads, but there is limited data from both countries in terms of the impacts of these roads.

3 Target group

The construction of new roads using the 2+1 design is likely to benefit all road users with the reduction of head-on collisions. For those road users who cause the possibility of accident, for instance by sleeping behind the wheel, the immediate benefit is the reduction in the probability to have a serious head-on collision. For drivers who are in danger of facing a possible head-on collision, the benefit is derived from the fact that the probability again is diminished by the existence of a middle road cable.

4 Assessment method

Sweden has been ahead of Finland in designing and implementing the 2+1 road construction. There is a relatively well-documented, existing database on the constructed sections of the 2+1 roads, produced by VTI, the Swedish National Institute for Transport Research. In fact, VTI has been responsible for annual follow-up studies on the 2+1 roads (or, in more general terms, the head-on collision free roads). The unit cost for one kilometre of 2+1 road in Sweden was estimated at € 125,000, but since actual costs of the investment were available, the real figures were used instead.

Table 74: Accident costs, official values [SEK]

TYPE OF MATERIAL COSTS RISK VALUE TOTAL ACCIDENT DEATH 1,242,000 16,269,000 17,511,000 SEVERE INJURY 621,000 2,503,000 3,124,000 SLIGHT INJURY 62,000 113,000 175,000 PROPERTY 13,000 13,000 DAMAGE

5 Method of analysis

The calculations have been complicated by the fact that the construction of road sections in Sweden takes place on the terms of developing the road as a whole, as opposed to constructing a separate measure, such as the pavement alone. In fact, in many cases the centre of the road pavement is part of an upgrading of existing road, where traffic volumes have increased to degrade the existing road. We do not want to separate the safety effect in the analyses (for instance, in the form of cost-effectiveness analyses of various safety measures, as this is not the procedure applied in the national project appraisal. The Swedish case is from RV 44, Trollhättan-Håsten, totalling 10.6 km. The road was opened as a typical 13-meter, 2-lane road in 1990. Daily traffic volume between 1991-99 was calculated to be 6450 vehicles. In 2000 the upgrading of the road began with installation of the mid-road cable and the new road consisted of 6 sections of 2+1 road,

Page 219 2 +1 ROADS IN SWEDEN stretching from 910 meters to 1880 meters. Total cost of the operation was 44.6 million Swedish Kronor (5 million €). On this road section the average cost per kilometre was higher than the average estimate of 125,000 Euros (nearly 500,000 € per kilometre). Accident statistics for the road show that during the 2+1 solution there were eight reported accidents for the period of first 18 months of the operation of the new road, with two person accidents (slight injuries). These accidents were used to correct the accidents data that would consider the reduction of deadly accidents. As in the Finnish case, similarly we need to calculate: • Accident costs • Time savings • Vehicle costs • Emissions • Noise • Maintenance costs • Investments For several of the variables averages were used based on the traffic volumes. This is because the real data had not been collected for the purposes of the socio-economic profitability, or if such data existed, it was not available for this case study. The next section presents the results of the calculations.

6 Assessment Results

Assessment was carried out using the socio-economic profitability analysis, which is the standard method of road investment project assessment in Sweden. Carrying out the calculations for the project (with a standard duration of 20 years and a 4 percent discount rate) gives us the CBA results in the form of socio-economic profitability with all the mentioned elements of the analysis. For the case of RV 44, Trollhättan-Håsten, the calculations yield a benefit-cost ratio of 2.26. The ratio is good, making the project profitable. The main sources of benefits were derived from safety impact (reductions in estimated deaths) and time savings due to the overtaking lane.

7 Decision-Making Process

In 1998 the Director General of the Swedish National Road Authority decided on a full- scale programme to improve traffic safety on six existing 13-meter roads using low-cost measures, where the main alternative identified was the 2+1 road with the separating median cable. The estimate was to have a potential to reduce 50 percent of all severe link accidents. It has been thereafter indicated that all old 13-meter roads should be replaced with the 2+1 roads. In the Swedish system, the road administration (Vägverket) produces and executes the investment plans. The final decision-making authority is in the hands of the parliament, which confirms the annual budget for road construction. The parliament, which is committed to the Swedish zero vision (on traffic deaths) has clearly followed the principle in promoting the 2+1 road and other non-collision construction methods for new roads. The political atmosphere is therefore clearly favourable to implement safety-improving measures.

Page 220 2 +1 ROADS IN SWEDEN

8 Role of barriers

Barriers in Sweden tend to be similar to those reported in the Finnish case, but more appearing as a result of financial constraints than simply those of political nature. It appears that several interest groups have been active in promoting the 2+1 road as one of the major tools in reducing traffic accidents in Sweden. Perhaps in Sweden the relatively low cost of this measure can be better understood as an alternative to motorways in the areas where the traffic volumes do not suggest that a motorway is required to remove capacity bottlenecks. As in the Finnish case, data was relatively easily available and the quality was satisfactory. Earlier studies of VTI had focused on more traffic flows than economic assessment, so there was need to supplement the basic data with data on investment costs. These additional data requirements did not complicate the analyses.

9 Discussion

In Sweden, the benefits of constructing the 2+1 road have been clearly documented well in advance. The public opinion has been in favour of the solution, as it is considered an effective means of preventing head-on collisions and is cost-effective compared to motorway construction. More than in Finland, in Sweden the 2+1 road construction is understood as a safety measure, but this is not the principal criteria for constructing the road. Like in Finland, also in Sweden the socio-economic profitability approach dominates cost-effectiveness approach. Perhaps the biggest challenge for shifting towards consideration of specific measures and their appraisal is to acknowledge that decision-making can take place on the basis of, for instance, the cost-effectiveness of the measure. The realization that not all the projects can be comparable, if they are based on a single target (e.g., safety) compared to multiple targets, which could include mobility, time savings and safety.

REFERENCES

CARLSSON, ARNE et al. (2003): Uppföljning av mötesfria vägar. Halvårsrapport 2002:1. VTI notat 9-2003. CARLSSON, ARNE and ULF BRÜDE (2003): Utvärdering av mötesfri väg. Halvårsrapport 2002:2. VTI notat 45-2003. LARSSON, M., T. BERGH JA A. CARLSSON (2003): Swedish Vision Zero Experience. NTF (2003): Motorvägar dödar fler än de räddar. NTF Tidning.

Page 221 CASE K: compulsory bicycle helmet wearing

ROSEBUD WP4 - CASE K REPORT

COMPULSORY BICYCLE HELMET WEARING

BY MARTIN WINKELBAUER,

AUSTRIAN ROAD SAFETY BOARD, KFV, AUSTRIA COMPULSORY HELMET WEARING FOR CYCLISTS

TABLE OF CONTENTS

1 EFFICIENCY ASSESSMENT FOR GERMANY ...... 225 1.1 Problem to solve ...... 225 1.2 Description ...... 225 1.3 Target Group...... 225 1.4 Assessment method...... 225 1.5 Choice of Efficiency Assessment method ...... 225 1.6 Assessment tool and calculation method ...... 226 1.6.1 Types of assessed impacts: safety, environment, mobility, travel time ...... 226 1.6.2 Considered cost of the measure ...... 226 1.7 Assessment Quantification...... 227 1.7.1 Target group...... 227 1.7.2 Current helmet wearing rates...... 227 1.7.3 Accident statistics...... 227 1.7.4 Helmet prices ...... 228 1.7.5 Accident reduction potential ...... 228 1.7.6 Crash costs ...... 229 1.7.7 Unit of Implementation ...... 229 1.7.8 Price basis, interest rates and duration of the measure ...... 229 1.8 Assessment Results...... 229 1.8.1 Calculation procedure ...... 229 1.8.2 Cost-benefit ratio by expected values ...... 230 1.8.3 Marginal cost-effective helmet wearing rates ...... 230 1.9 Decision Making Process...... 231 2 EFFICIENCY ASSESSMENT FOR AUSTRIA...... 231 2.1 Problem to solve ...... 231 2.2 Description ...... 232 2.3 Target Group...... 232 2.4 Assessment method...... 232 2.4.1 Assessment tool and calculation method ...... 232 2.4.2 Types of assessed impacts: safety, environment, mobility, travel time ...... 232 2.4.3 Considered cost of the measure ...... 233 2.5 Assessment Quantification...... 233 2.5.1 Target group...... 233 2.5.2 Current helmet wearing rates...... 234 2.5.3 Accident statistics...... 234 2.5.4 Helmet prices ...... 235 2.5.5 Accident reduction potential ...... 235 2.6 Assessment Results...... 236 3 DECISION MAKING PROCESS...... 237 4 IMPLEMENTATION BARRIERS ...... 237 5 CONCLUSION / DISCUSSION...... 238

Page 223 COMPULSORY HELMET WEARING FOR CYCLISTS

CASE OVERVIEW

Measure Compulsory bicycle helmet wearing Problem Among all severe injuries sustained by bicycle riders, head injuries are the most common. At the same time, average helmet wearing rates are very low. The protective potential of a bicycle helmet is considered to be very high. Target Group All bicycle riders (precisely those currently not wearing a helmet) Targets Reduction of head injuries among bicycle riders Initiator Research institutes Decision-makers The decision has to be made by the national parliaments and has to be prepared following the usual procedures for such legislation. Costs Helmet costs Benefits Reduction of head injuries and all related costs Cost-Benefit Ratio

efficiency of compulsory helmet cost/benefit ratio wearing Germany Austria road accidents only all accidents helmet price € 20 4.45 2.28 4.10 € 40 2.23 1.14 2.05

Page 224 COMPULSORY HELMET WEARING FOR CYCLISTS

1 Efficiency Assessment for Germany

1.1 Problem

62% of the German population use a bicycle at least occasionally [Mobilität in Deutschland 2002 – Fahrradverkehr]. Annually, about 600 (in 2003: 639) Germans are killed as bicyclists in road traffic, about 15,000 (in 2003: 15,591) are severely injured and 65,000 slightly injured. A little less than 50% of the bicyclists injured in road traffic suffer head injuries. 65% of the head injuries occur in regions of the head that are covered by a helmet and therefore are potentially protected by helmet wearing. In total, about 20% of the fatal and severe injuries may be avoided by helmet wearing and the number of slight injuries will rise by 1% if all bicyclists would wear helmets [OTTE, 2001]. Although the safety potential of wearing a cycle helmet is high and well documented, helmet wearing rates are still very low. A considerable share of children wear helmets (about 60%); the average helmet wearing rate in Germany is almost constant over the recent years, currently about 6% [SIEGENER, 2004]. Bicycle helmet wearing campaigns have been carried out successfully, but the total wearing rate could not be raised to a desirable level.

1.2 Description

To make bicycle helmet wearing compulsory for all bicyclists. Used helmets shall be approved by using one of the existing standards for cycle helmets (e.g. EN 1078).

1.3 Target Group

The target group is those bicyclists currently not wearing a helmet, which is a huge majority of bicyclists in Germany.

1.4 Assessment method

1.5 Choice of Efficiency Assessment method

It was decided to perform a cost-benefit analysis for the following reasons: • It was an explicit demand of the partners in Germany (bast) to choose CBA. • The potential of injury reduction is well documented, but it did not support the decision- making process in a satisfying manner. The question of cost benefit in relation to the public economy level was raised during this process. • Most of the fatalities and severe injuries are considered to remain as slight injuries after introducing the measure, while the effect of helmet wearing on slight injuries is small. This leads to differing impacts of the measure on different levels of injury severity, which cannot be considered in a CEA.

Page 225 COMPULSORY HELMET WEARING FOR CYCLISTS

There was no question of comparing helmet wearing with other safety measures, in particular measures dedicated to bicyclists (for which a CEA would have been useful). As indicated, the effectiveness of helmet wearing is not in doubt at all, a cost- effectiveness ratio would not have given any severe input to the decision-making process.

1.6 Assessment tool and calculation method

A self-made calculation method was chosen using a spreadsheet program. The main inputs were accident and population data and helmet wearing rates. Both were available in age groups and for several years. Partly, the data was aggregated to age groups with different thresholds. There were big differences between age groups. It seemed easy to calculate the data without using formal assessment methods.

1.6.1 Types of assessed impacts: safety, environment, mobility, travel time

Safety Concerning safety, three effects may be considered: 1. Reduced likeliness of head injury 2. Increased risk by risk compensation 3. Reduced risk by reduced exposure It was decided not to consider risk compensation and changes of exposure for the following reasons: • Emotionally based effects like risk compensation and change of mobility behaviour are very much based on the culture in the target country. There was no evidence that these effects should occur in Germany. • It was presumed, that those who object to wearing a helmet would not change their mode of mobility, but continue cycling without a helmet. This is why a "break even helmet wearing rate" was calculated afterwards. • It was also presumed, that there may be a group of bicyclists who take higher risks if wearing a helmet. But a majority of these may be found among the bicyclists already wearing a helmet. Those cyclists who wear helmets due to legal obligation were not assumed to change their risk behaviour significantly.

Environment, mobility, travel time As indicated above, a significant change of modal split was not expected to occur. If this is the case, there will be no significant impact on environment, mobility and travel time.

1.6.2 Considered cost of the measure

The costs of the measure simply consist of the costs for supplying bicycle riders with helmets. The cost of the legal process (making the law) will not be considered. Due to the Page 226 COMPULSORY HELMET WEARING FOR CYCLISTS decision not to consider effects of a modal shift, no costs of environmental effects, mobility or travel time will be considered. The time use for the handling of the helmet was considered to have a very low impact on total travel time and was therefore disregarded.

1.7 Assessment Quantification

1.7.1 Target group

The definition and calculation of the target group was primarily based on the total population. "Mobilität in Deutschland 2002 -Fahrradverkehr" presents data on the frequency of bicycle use; 38% of the Germans never use a bike and were discounted. Cyclists already wearing a helmet had to be excluded from the calculation. There are no impacts from this group either on accidents (the accident statistics and their development already represented the impact of them wearing a helmet) or on costs (the money for their helmets was already spent and a helmet law will have no impact on replacement costs of these helmets).

1.7.2 Current helmet wearing rates

Table 75: Helmet wearing rates in Germany Helmet wearing rates age groups Germany - 5 6 - 10 11 - 17 - 22 - 31 - 41 - > 60 total 16 21 30 40 60 year of 1997 59% 37% 12% 3% 3% 3% 2% 1% 6% assessment 1999 85% 47% 11% 2% 3% 3% 2% 1% 5% 2001 58% 37% 8% 2% 3% 3% 3% 1% 5% 2002 32% 33% 9% 2% 3% 4% 3% 2% 5% 2003 60% 38% 10% 2% 2% 5% 5% 2% 6%

The study on helmet wearing rates by Siegener (2004) is based on a sample of 6800 to 8300 observations in each of the years. The sample of children is very small (32-80 observations) and therefore not very reliable. But the figures were compared to a study from Austria with a larger sample and found plausible.

1.7.3 Accident statistics

The German accident data contains road accidents taken from the official accident database including the years from 1991 to 2003 (disaggregated data from the German in- depth-analysis-system GIDAS combined with aggregated). This database contains all injuries where any of the parties involved sustained personal injury. It also contains the numbers of all bicyclists, killed, severely injured or slightly injured in road accidents. It does not contain the numbers of bicyclists killed or injured apart from road traffic. Further, this database does not contain accidents that were not noticed by the police, i.e., all those cases where a bicyclist falls off the bicycle for any reason in a single party accident and

Page 227 COMPULSORY HELMET WEARING FOR CYCLISTS goes away injured without calling the police are not contained. The population used was taken from the official population statistics. For Germany, there was good information on vehicle numbers available. This study by the Deutsches Institut für Wirtschaftsforschung (DIW), Berlin, does not contain bicycles qualified as children's toys. This data later on was not used for calculation, but kept in this report for information purposes. Unfortunately, the age classification in the different data sources differs from each other. Punctually, age classes had to be summarised together or divided based on the population data. The whole table of German data can be found in annex K1.

1.7.4 Helmet prices

The prices of bicycle helmets differ very much. The cheapest offers are available for children's helmets in super-markets, which are about € 7. The most expensive helmets are about € 95. Elvik (2004) indicates helmet prices for children are about € 35 to € 50, an adult helmet between € 50 and € 62. He estimates the lifetime of a children's helmet about 3 years, a young adult helmet about 6 years and an adult helmet about 10 years. In the USA and Australia the average helmet price is about € 25 to € 30. A short investigation in Austria (currently no data available either in Austria or in Germany) had poor results, as most of the companies only gave little information on the prices of helmets, and what would have been necessary to weight this data, not any information on their sales or market share. The only really useful information came from a big sport supplier, telling us that the average price of a cycle helmet is slightly below € 40. What had to be taken into consideration was the number of helmets sold, if helmet wearing is compulsory. For example, rescue jackets (warning jackets) were available in Austria for about € 15. Immediately after introducing a law that rescue jackets will be compulsory, even before this law was put into force, the prices fell to € 4. It is supposed that a similar effect will take place if cycle helmet wearing should become compulsory. Further, we can suppose that people buying a helmet voluntary for their own safety have different patterns of decisions in their helmet purchase than those buying a helmet due to a legal obligation. Determination of a suitable price for helmets is a key issue in this CBA since helmet costs are the only cost factor. To consider the uncertainty of future helmet prices, it was decided to calculate two alternatives: A conservative one with a helmet price of € 40 (i.e. helmets at current price level) and a progressive one with € 20 as the average price for a helmet.

1.7.5 Accident reduction potential

There were various studies on the injury reduction potential of bicycle helmets. As it was most commonly accepted in Germany, a study by OTTE (2001) was chosen as reference for this assessment. OTTE investigated in-depth 3534 accidents with bicyclists involved in Germany between 1985 and 1999. This very elaborate study considers injuries of different regions of the body and different regions of the head. Actual injury severity of real life crashes is compared to virtual injury severity, i.e. injuries which would have occurred if a helmet had worn. OTTE concludes that the total number of fatally and seriously injured bicycle riders would decline by 20% if all cyclists would wear helmets. The number of slight injuries would increase by 1%, since the number of slight head injuries protected by a helmet is lower than the number of fatal and severe injuries changed to slight ones.

Page 228 COMPULSORY HELMET WEARING FOR CYCLISTS

1.7.6 Crash costs

The accident costs (fatalities, severe and slight injuries) are taken from the ROSESUD WP3 report for Germany, which is the official German accident cost estimation.

1.7.7 Unit of Implementation

Compulsory cycle helmet wearing is a measure that applies to all cyclists in the whole country. The crash reduction potential is estimated for a whole country. The decision has to be made for the whole country. Finally it is presumed that the results of an assessment would be most useful if they estimate the total effect in reference to the group that is affected by the measure, which is a whole country again. So it was decided to consider the whole country as the unit of implementation.

1.7.8 Price basis, interest rates and duration of the measure

For all values presented in the previous ROSEBUD deliverables, it was decided for comparability reasons to convert all monetary values to 2002 prices. It was then agreed to choose the same procedure also for WP4 cases. The interest rate was chosen based on ROSEBUD WP3 recommendations. The life span of a cycle helmet was considered between 3 and 10 years. This recommends assessing a period of at least 10 years. ROSEBUD WP3 recommends assessing a period of 20 to 30 years, for non-infrastructure measures the period may be shorter. Based on that, it was decided to assess a period of 13 years, being somewhere in between 10 and 20 years.

1.8 Assessment Results

1.8.1 Calculation procedure

• Based on the reported accident data (1991 to 2002), forecasts for 2003 to 2015 were calculated and reviewed for plausibility. • The same was done for helmet wearing rates based on the data from 1997 to 2003. Wearing rates for 1998 were missing (not investigated) and were interpolated. • The target accidents affected by a helmet wearing obligation was calculated as the product of the share of cyclists currently not wearing a helmet and the number of injuries in the different levels and age groups. • A further calculation was done as if the helmet law would have been introduced on January 1st 2003. • The crash severity reduction figures were applied to the target accidents in the different age groups and injury severity classes.

Page 229 COMPULSORY HELMET WEARING FOR CYCLISTS

• Future costs and benefits were labelled to 2002 prices using a discount factor of 5% annually. • The costs were calculated assuming that 38% of the Germans never use a bike and the rest will be fully equipped with helmets. • Afterwards, two approaches were chosen.

1.8.2 Cost-benefit ratio by expected values

Considering the predictions for accidents, helmet wearing rates, population and accident reduction potential, the costs and benefits were calculated for two values of the expected helmet price. Within the period from 2003 to 2015 the cumulated costs and benefits based on 2002 prices will be: Table 76: Costs and benefits 2003 -2015, Germany

Helmet price benefits total supply costs cost/benefit-ratio (€) (€) (€)

20,- 5,077,319,223 1,140,167,632 4.45

40,- 5,077,319,223 2,280,335,263 2.23

1.8.3 Marginal cost-effective helmet wearing rates

For this presentation of the result it was supposed that 100% of the Germans at least occasionally riding a bike buy a helmet. Again, supposing two different prices of the average helmet, the minimum helmet wearing rate which would achieve a cost-benefit ratio of one was calculated, i.e. enforcement measures would have to achieve at least a "break-even helmet wearing rate" of 26.6% (47.9%) to make bicycle helmets effective supposing a worst case scenario for the costs. Table 77: marginal average helmet wearing rates 2003- 2015, Germany

Helmet price benefits total supply costs break-even (€) (€) (€) helmet wearing rate

20,- 1,140,464,390 1,140,167,632 26.6%

40,- 2,282,903,190 2,280,335,263 47.9%

There is one problem in this type of calculation, as there is no information available on enforcement costs either on costs of one unit of enforcement (e.g. one hour of road-side enforcement) or on the number of those units necessary to achieve the demanded helmet wearing rate. Further it is not known whether these enforcement measures would be self- financing by fines. Even if enforcement measures would be cost neutral for the authorities, they might not be for the target group.

Page 230 COMPULSORY HELMET WEARING FOR CYCLISTS

1.9 Decision-Making Process

• Currently there is no governmental initiative for making helmet wearing compulsory in Germany. But there is an explicit backup and encouragement for voluntary helmet wearing. • Basically the government strongly aims at improving road safety and reducing accident costs. Compulsory helmet wearing fits into this target but is currently not at a status of official discussion. • The bicycle-rider lobby wants to avoid any interference in bicycling, referring to aspects like "comfort", "freedom" and "responsibility". These groups argue that a large share of cyclists would stop cycling if a helmet would have to be worn. • A decision about an obligation to wear a helmet would have to be made by the national parliament, and the German Diet "Deutscher Bundestag". • It was supposed by a member of the ROSEBUD URG to select compulsory bicycle helmet wearing as one of the ROSEBUD WP4 cases. • But due to the fact that currently there was no occasion to raise a political or public discussion about compulsory helmet wearing, the results had only been presented to decision-makers from inside the experts' organisation. These experts agreed to the findings, but they did not think that EA results would bring useful input to the process of political and public decision making. The case was too far away from being discussed on a rational basis, that monetary arguments at national level, which are hardly understandable for the public, could support the implementation of this measure. • So far, it cannot be foreseen when a public and political discussion on bicycle helmet wearing will be continued.

2 Efficiency Assessment for Austria

2.1 Problem to solve

62% of the Austrian population uses a bicycle at least occasionally [BÄSSLER, 2001]. Annually, about 60 (in 2003: 56) Austrians are killed as bicyclists in road traffic, about 1,800 (in 2003: 1,838) are severely injured and 4,000 slightly injured. A little less than 50% of the bicyclists injured in road traffic suffer head injuries. The Austrian Federal Ministry of Transportation, Innovation and Technology has set up a Road Safety Program from 2002 to 2010 which includes a 50% reduction target for fatalities. Although this program does not specifically mention cycle helmet wearing as a measure targeting bicycle accidents, cycle helmet could achieve a serious contribution towards road safety targets. Besides, there might be an additional contribution in reducing injury severity after leisure time and sport accidents. Although the safety potential of cycle helmet is high and well documented, the helmet wearing rates are currently very low. A considerable amount of children wear helmets Page 231 COMPULSORY HELMET WEARING FOR CYCLISTS

(about 60%), the average helmet wearing rate in Germany is almost constant over the recent years, currently about 11% [FURIAN, GRUBER, 2002]. Bicycle helmet wearing campaigns have been carried out successfully, particularly targeting school children, but the total wearing rates could not be raised to a desirable level neither among children nor among adults.

2.2 Description

Making bicycle helmet wearing compulsory for all bicyclists. Used helmets shall be approved by using one of the existing standards for cycle helmets (e.g. EN 1078).

2.3 Target Group

Those bicyclists currently not wearing a helmet, which is a huge majority of bicyclists in Austria.

2.4 Assessment method

It was decided to perform a cost-benefit analysis (CBA) for the following reasons: • The studies for Germany and Austria were done at the same time, for reason of comparability it was useful to select the same method. • Most of the fatalities and severe injuries are considered to remain as slight injuries after introducing the measure, while the effect of helmet wearing on slight injuries is small. This leads to differing impact on different levels of injury severity, which cannot be considered in a CEA. There was no question of comparing helmet wearing with other safety measures, in particular measures dedicated to bicyclists (for which a CEA would have been useful). As indicated, the effectiveness of helmet wearing is not in doubt at all; a cost- effectiveness ratio would not have given any severe input to the decision-making process.

2.4.1 Assessment tool and calculation method

A self-made calculation method was chosen using a spreadsheet program. The main inputs were accident and population data and helmet wearing rates. Both were available in age groups and for several years. Partly, the data was aggregated to age groups with different thresholds. There were considerable differences between age groups. It seemed easy to calculate the data without using formal assessment methods.

2.4.2 Types of assessed impacts: safety, environment, mobility, travel time

Safety Concerning safety, three effects may be considered:

Page 232 COMPULSORY HELMET WEARING FOR CYCLISTS

• Reduced likeliness of head injury • Increased risk by risk compensation • Reduced risk by reduced exposure It was decided not to consider risk compensation and changes of exposure for the following reasons: • Emotionally based effects like risk compensation and change of mobility behaviour are very much based on the culture in the target country. There was no evidence that these effects should occur in Austria. • It was presumed that those who object wearing a helmet would not change their mode of mobility, but continue cycling without a helmet. • It was also presumed that there may be a group of bicyclists who take higher risks if wearing a helmet. But a majority of these may be found among the bicyclists already wearing a helmet. Those cyclists who wear helmets due to legal obligation were not assumed to change their risk behaviour significantly. Environment, mobility, travel time As indicated above, a significant change of modal split was not expected to occur. If that is the case, there will be no significant impact on environment, mobility and travel time.

2.4.3 Considered cost of the measure

The costs of the measure simply consisted of the costs for supplying bicycle riders with helmets. The costs of the legal process (making the law) were not considered. Due to the decision not to consider effects of a modal shift, no costs of environmental effects, mobility or travel time had to be considered. The time use for the handling of the helmet was considered to have a very low impact on total travel time and was therefore disregarded.

2.5 Assessment Quantification

2.5.1 Target group

The definition and calculation of the target group was primarily based on the total population. Baessler (2001) showed numbers of inhabitants at least occasionally performing various sports; 4.1 million Austrians aged over 15 engage in cycling. Extrapolating this value to persons under 15 considering the share of the total population gave almost exactly the same figures as in Germany. Persons already wearing a helmet had to be excluded from the calculation. There will be no impacts from this group either on accidents (the accident statistics and their development already represented the impact of them wearing a helmet) or on costs (the money for their helmets is already spent and a helmet law will have no impact on replacement costs of these helmets).

Page 233 COMPULSORY HELMET WEARING FOR CYCLISTS

2.5.2 Current helmet wearing rates

Table 78: helmet wearing rates in Austria helmet wearing rates year of assessment Austria 1992 1994 1996 1998 2001 total 2.7% 5.7% 8.6% 11.4% 10.7% by age children 5.6% 19.3% 28.2% 42.9% 43.3% juvenile 1.7% 5.2% 7.0% 12.5% 8.2% adult 2.5% 4.4% 7.0% 8.1% 8.9% by mode of sports 6.0% 16.0% 18.0% 22.0% 30.0% bicycle use leisure time 1.0% 3.0% 4.0% 9.0% 7.0% traffic 3.0% 3.0% 7.0% 7.0% 7.0% by type of children 6.0% 25.0% 36.0% 59.0% 63.0% bicycle adult bike 1.0% 2.0% 3.0% 4.0% 5.0% mountain bike 3.0% 7.0% 11.0% 12.0% 12.0% street race 13.0% 17.0% 20.0% 44.0% 62.0% bike

These results of a study by Furian and Gruber (2001, 2002 and 2003) are based on about 20,000 observations in each of the years, which should provide very reliable information. But the observations were made on bicyclists passing by without asking them for their age. So the age distribution between children, juveniles and adults is only based on the estimation of the observers. A study on helmet wearing rates in 2004 is currently being carried out, but results were not available. Generally speaking, the helmet wearing rates changed extremely over the years, which makes forecasts quite difficult.

2.5.3 Accident statistics

Bicycle accidents are divided into two groups: • Road traffic accidents, i.e. accidents occurring on public roads taken from the official road traffic accident database. • The EHLASS database provides accident data based on about 12,000 interviews annually. These interviews are carried out in hospitals with interviewees who have sustained leisure time accidents. The accident statistics of the "Institut Sicher Leben" summarises sport and leisure time accidents. It may be presumed that these accidents are separate from traffic accidents, but no data on fatalities is included and the accident severity is reported in other patterns than the traffic accidents. This database contains data on injuries of different regions of the body, including head injuries. The evaluation has to deal with shortcomings in both of these sources: • It is supposed that there are a considerable number of unreported cases not contained in the official traffic accident database, e.g. single party accidents of cyclists. Due to a

Page 234 COMPULSORY HELMET WEARING FOR CYCLISTS

legal obligation all road accidents where any of the persons involved sustains any injury have to be reported by the police. But particularly cycle accidents are not likely to be reported if no other party is involved. However, it is likely that most accidents remaining unreported this way are of minor severity and therefore will not alter the result of a CBA significantly. • Injuries not treated in hospitals, but by general practitioners in their private practices, are not covered in either database. The same for injuries that are not at all treated by doctors, however, these injuries may be frequent but are assumed severe enough to have a significant impact on public economy. • Fatal leisure time accidents are not reported. • Since the EHLASS data is investigated by interviews, it likely but not secure that no accidents are double-counted in both databases. • In the EHLASS database the accident severity is reported in other patterns than the traffic accidents. Unfortunately the age classification thresholds in the different data sources differ from each other. Punctually and age classes had to be summarised together or divided based on the population data. This data was used to calculate two different scenarios, one for road traffic on and one for all accidents. A comparison of both databases shows that on an average road traffic accidents are much more severe than sport and leisure time accidents. This difference was considered in the calculation. Sport and leisure time accident data was only available for 2001 to 2003; the numbers differ significantly. There are only three age groups.

2.5.4 Helmet prices

For the Austrian study the same approach was used as for Germany.

2.5.5 Accident reduction potential

There are various studies on the injury reduction potential of bicycle helmets. As it is most commonly accepted in Germany, a study by OTTE (2001) was chosen as reference for this calculation. OTTE investigated in-depth 3,534 accidents with bicyclists involved in Germany between 1985 and 1999. This very elaborate study considers injuries of different regions of the body as a whole and for different regions of the head. Actual injury severity of real life crashes is compared to virtual injury severity (i.e. injuries which would have occurred if a helmet would have been worn). It comes to the final conclusion that the total number of fatally and seriously injured bicycle riders would decline by 20% if all cyclists would wear helmets. The number of slight injuries would increase by 1% since the number of slight head injuries protected by a helmet is lower than the number of fatal and severe injuries changed to slight ones.

Page 235 COMPULSORY HELMET WEARING FOR CYCLISTS

2.6 Assessment Results

• Based on accident data of 1992 to 2003, forecasts for 2004 to 2015 were calculated. • The further calculation was done as if the helmet law would have been introduced on January 1st 2003. • The data on helmet wearing rates available does not allow one to extrapolate wearing rates for the future; there was a rapid increase in the 90s, whereas the wearing rates declined slightly recently. Therefore the helmet wearing rates of 2001 were taken as the basis to define the target group by excluding the share of cyclists currently wearing a helmet. • The target accidents affected by a helmet wearing obligation were calculated as the product of the share of those not wearing a helmet and the number of injuries in the different levels and age groups. • The crash severity reduction figures were applied to the target accidents in the different age groups and injury severity classes. • The costs were calculated assuming that 62% of the Austrians, i.e. all those at least occasionally using a bike, buy a helmet. The information about the life span of helmets was taken from Rune Elvik's Handbook of Road Safety Measures. Regular re- investment for helmets was considered from this source. For discounting those already wearing a helmet the same figures were used as described above for accidents. • Future costs and benefits were labelled at 2002 prices using a discount factor of 5% annually, and then finally added up. • To integrate sport and leisure time accidents, the two databases had to be compared. Sport and leisure time accident numbers are only available for the years 2001 to 2003, and besides, the numbers vary significantly through the years. It was supposed that the impact of helmet wearing on sport and leisure time accidents and road accidents is the same. An accident ratio was calculated based on 2001 to 2003 values for the three age groups. This ratio consists of two factors. One considers the ratio of the total numbers of accidents; the other considers the share of head injuries being different for road and other accidents. Table 79: costs and benefits 2003 - 2015, Austria

helmet cost (€) benefits (€) total supply costs cost/benefit ratio (€) only road 20 230,918,822 101,081,159 2.28 accidents 40 230,918,822 202,162,319 1.14 all accidents 20 414,093,300 101,081,159 4.10 40 414,093,300 202,162,319 2.05

Page 236 COMPULSORY HELMET WEARING FOR CYCLISTS

3 Decision-Making Process

If helmet wearing for cyclists would be made compulsory, this would have to be a decision by the national parliament. Depending on what this obligation should cover, road traffic legislation would not only be necessary. Deriving from this fact, it would also be necessary to consult more than one ministry. After making an arrangement between the concerned ministries, a draft of the law would have to be sent out to be commented on by a lot of stakeholders. Usually this phase gives rise to the public discussion. As we have learned from informal consultations with other stakeholders during this operation, it may be expected that most of the interest groups would oppose to an obligation. The "Institut Sicher Leben" is a research institute dealing with sport and leisure time safety. This study was presented to the head of the institute and three experienced researchers working for the institute. The presentation of the study was started with presenting the "short training course" on efficiency assessment although some of the audience already had experience in this field. The "short training course" and the study itself were understood by the audience. The results were accepted. Nevertheless, the head of the institute decided not to bring the study forward to the relevant members of the Austrian administration. Well informed about the current positions of the stakeholders, he judged that it would do no good to the case itself if the discussion would be raised under the current circumstances. It was considered to present the efficiency assessment results to relevant decision-makers without joining this with a recommendation to implement the measure, or even pointing out that the institute does not recommend mandatory helmet wearing. But all these options were rejected as it seemed impossible to leave the discussion, only to leave it at a strictly theoretical level. But, as a positive result of this presentation, it was decided to prepare another CBA only considering children. The protection of children would not be the subject of a great deal of controversy as cycle helmet for all cyclists would be, safety measures for children cannot be easily objected, at least not as easy as measures targeting the whole population.

4 Implementation barriers

None of the fundamental barriers played a significant role within this study. The institutional barriers were finally those avoiding this study to be used as it was meant to be. It might have been a matter of wrong timing, but definitely was not the wrong timing of using EA results to influence decision-making. Since there was no decision-making process running, the results would have to be used to start a process. And finally, it was supposed that these EA results were not a good starting point for a political discussion. Within the calculation procedure several difficulties occurred, but the results of the previous work packages of ROSEBUD were found to be very helpful to overcome these problems. The valuation of fatalities and injuries used for this study significantly differs from, e.g. the values used in Germany. New values for Austria will be available by the end of 2005. Three main problems were identified in the range of technical barriers:

Page 237 COMPULSORY HELMET WEARING FOR CYCLISTS

• Accident data: Road traffic and off-road accident data were difficult to compare and to aggregate. Unreported accidents were presumed to exist in a considerable number, but having no considerable impact on the total accident costs. • The basis for the estimate of helmet prices was rather weak. Finally, there is no evidence of people stopping cycling when forced to wear a helmet. Although there are some results from another country tackling this problem, it seems to be highly depending on culture and attitudes of cyclists whether there is any change in mobility behaviour. If these effects should occur, there would be impacts of various kinds that would have to be considered in a CBA additionally (environmental effects and travelling time considering changed modal split as well as public health effects).

5 Conclusion / Discussion

When this study was carried out, there was strong interest by research institutes in EA of compulsory helmet wearing, but there was no discussion going on either among scientists and the administration or in the public. EA was found not to be an appropriate means of raising this discussion. It was supposed that the argument of cost efficiency would not be heard by the public, particularly not in a case where emotional arguments are the main basis of these discussions. It was found that it is not possible to discuss rational arguments (like the results of a EA study) with the relevant stakeholders on a broad basis without starting a public discussion on the topic at the same time. The limitations of the CBA carried out were identified as follows: • The accident data used may be influenced by a large number of unreported cases. • Accident data from road traffic accidents and another database containing leisure time accidents were difficult to compare and aggregate to a common basis for calculation. • Helmet prices were difficult to estimate due to a lack of knowledge on current prices and strong uncertainty of the impact of the enormous increase of sales after introducing an obligation. • The valuations for fatalities and injuries for Austria are based on relatively old data. • Although there is no evidence for these effects to exist, a change of the modal split would significantly alter the results of this CBA. Time consumption, environmental impacts and public health effects would have to be considered in addition. The following task in this CBA were particularly easy to achieve: • It was easy to gain access to accident data and other data sources, such as population data and empirical data on attitudes, mobility behaviour and helmet wearing rates. • There was suitable and elaborate information on the safety effects. • The calculation itself was supported by the framework described in WP3 report.

Page 238 COMPULSORY HELMET WEARING FOR CYCLISTS

An obligation for cyclists to wear a helmet was found beneficial in any case. A cost-benefit ratio was found between 1.14 and 4.45 depending on what accident types are included, the monetary values for fatalities and injuries and on the estimate for helmet prices.

REFERENCES

BÄSSLER, R. (2001): Quantifizierung des Unfallrisikos beim Sporttreiben. Austrian Life Style 2000. Fessel-GfK. Studie im Auftrag des Institutes "Sicher Leben". Austria. HALBWACHS C. et. Al. (2000): Sport und Gesundheit. Bundesministerium für soziale Sicherheit und Generationen. Wien. Austria. FURIAN G., Gruber M. (1999): Die Österreichische Radhelminitiative 1992 - 1998. Institut Sicher Leben. Wien. Austria. KOLB W., BAUER R. (1999): Unfallfolgekosten in Österreich. Institut Sicher Leben. Wien. Austria. STEINER M., BAUER R. (2002): Unfallstatistik 2001. Verletzte nach Heim-, Freizeit. und Sportunfällen in Österreich. Institut Sicher Leben. Wien. Austria. Steiner M., Bauer R. (2003): Unfallstatistik 2002. Verletzte nach Heim-, Freizeit. und Sportunfällen in Österreich. Institut Sicher Leben. Wien. Austria. BAUER R., KÖRMER, C., STEINER M. (2002). EHLASS Austria Jahresbericht 2001. Institut Sicher Leben. Wien. Austria. BAUER R. et al (2003):EHLASS Austria Jahresbericht 2002. Institut Sicher Leben. Wien. Austria. FURIAN G., Gruber M. (2002):Einstellungen zum Helmtragen, Verwendung von Radhelmen und Em,pfehlungen für die Zukunft. Institut Sicher Leben. Wien. Austria. OTTE, D. (2001): Schutzwirkung von Radhelmen. Verkehrsunfallforschung Medizinische Hochschule Hannover. Im Auftrage der Bundesanstalt für Straßenwesen. Bergisch Gladbach. Germany. N.N. (2004):Mobilität in Deutschland 2002 - Fahrradverkehr. Bundesministerium für Verkehr-, Bau- und Wohnungswesen. Bonn. Germany. SIEGENER W., RÖDELSTAB Th. (2004): Sicherung durch Gurte, Helme und andere Schutzsysteme. IVT Ingenieurbüro für Verkehrstechnik GmbH Karlsruhe. Bundesanstalt für Straßenwesen. Bergisch Gladbach. Germany. ELVIK, R., BORGER-MYSEN, A. and VAA, T. (1997): Trafikksikkerhekshandbok (Traffic Safety Handbook). Institute of Transport Economics. Oslo. Norway. ROSEBUD WP3 Report (2004): Improvements in efficiency assessment tools. ROSEBUD WP2 Report (2004): Barriers to the use of efficiency assessment tools in road safety policy.

Page 239 COMPULSORY HELMET WEARING FOR CYCLISTS

Accident, population and vehicle data, Germany, 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 injury accidents all 395462 385384 392754 388003 373082 380835 377257 395689 382949 375345 362054 with bicycle involved 78695 72487 74955 72949 66667 73341 68879 76133 73927 72110 71219 bicycle riders and total 925 906 821 825 751 594 679 637 662 659 635 583 616 passengers, fatalities 0-51212106126772444 6-10 37 28 33 33 24 27 20 18 25 10 12 10 10-15 59 58 74 66 54 45 39 48 53 41 37 28 15-21 62 68 43 59 57 40 36 33 35 40 39 36 21-65 437 430 377 371 326 277 316 293 289 298 278 270 >65 316 308 284 290 277 198 261 237 257 265 265 235 bicycle riders and total 17698 18928 17468 18041 17552 15747 17112 15624 16740 15586 14741 14025 passengers, severe 0-5 315 383 310 287 273 239 253 177 190 144 130 126 injuries 6-10 1393 1268 1198 1181 1241 1175 1088 873 942 717 532 490 10-15 2510 2704 2609 2657 2564 2290 2565 2134 2340 2014 1828 1606 15-21 2199 2401 2183 2263 2178 1823 1955 1773 1836 1601 1509 1481 21-65 8705 9586 8782 9061 8738 7748 8747 8188 8718 8325 8086 7618 >65 2550 2552 2358 2561 2529 2440 2484 2462 2692 2775 2646 2698 bicycle riders and total 52307 58552 53764 55507 54049 49647 54876 52053 58294 57152 56338 56138 passengers, slight 0-5 973 1075 958 910 922 742 805 634 744 652 591 615 injuries 6-10 3553 3582 3494 3427 3715 3431 3755 3068 3367 2838 2309 2258 10-15 8443 9378 9141 9207 8867 8323 9072 8414 9994 9152 8435 8420 15-21 7893 8814 7800 8061 7786 7044 7615 7505 7798 7464 7381 7296 21-65 27493 31394 28321 29689 28393 26003 28918 28015 31114 31315 31683 31420 >65 3719 3982 3770 3907 4051 3814 4351 4165 4953 5438 5681 5922 German population total 79984 80594 81179 81422 81661 81896 82052 82029 82087 82188 82339 82440 (x1000) 0-5 5357 5366 5319 5197 5051 4919 4832 4781 4743 4724 4706 4695 6-10 4211 4290 4377 4456 4517 4560 4569 4540 4506 4462 4396 4358 10-15 3445 3510 3582 3645 3695 3731 3738 3714 3687 3650 3596 3566 15-21 5363 5190 5103 5096 5177 5299 5411 5474 5521 5561 5590 5604 21-65 49640 50139 50526 50581 50586 50596 50587 50506 50421 50381 50177 50152 >65 11969 12100 12272 12448 12634 12791 12916 13014 13207 13510 13874 14066 Bicycles existing (million) 64,2 67,3 70 72,3 73,5 73,9 74 74 74,1 74,5 74,6

Page 240 SHORT TRAINING COURSE ON EFFICIENCY ASSESSMENT

SHORT TRAINING COURSE ON EFFICIENCY ASSESSMENT by Shalom Hakkert Efficiency assessment studies should be evaluated using standardized techniques. As highlighted at several points of this report, efficiency assessment is a sophisticated method and need some basic information to understand the methodology. This understanding is supposed as a basis for the recipients to believe in the results of such studies. The "short training course on efficiency assessment" provides a concise description of the main steps and data components, which are needed to perform a Cost-Benefit Analysis (CBA)/ Cost-Effectiveness Analysis (CEA) of a road safety measure25. The description includes: basic formulae, safety effects, implementation units, target accidents, accident costs and implementation costs. The evaluation of WP4 case-studies was performed in line with these evaluation techniques. Certainly, the background and interest of the recipients of efficiency assessment studies is very diverse. The "short training course" aims at making a compromise for all level of decision making and the full range of interests and background.

The introduction gives an overview on the motives to carry out EA studies, to use the results and the methods.

25 This is a concise compilation of Chapters 2, 3 of the WP3’s report. More details can be found in the report. Page 241 SHORT TRAINING COURSE ON EFFICIENCY ASSESSMENT a. Basic formulae The cost-effectiveness of a road safety measure is defined as the number of accidents prevented per unit cost of implementing the measure: Cost-effectiveness = Number of accidents prevented by a given measure/ Unit costs of implementation of measure For this calculation, the following information items are needed: • A definition of suitable units of implementation for the measure, • An estimate of the effectiveness of the safety measure in terms of the number of accidents it can be expected to prevent per unit implemented of the measure, • An estimate of the costs of implementing one unit of the measure. The accidents that are affected by a safety measure are referred to as target accidents. In order to estimate the number of accidents it can be expected to prevent (or prevented) per unit implemented of a safety measure, it is necessary to: • Identify target accidents, • Estimate the number of target accidents expected to occur per year for a typical unit of implementation, • Estimate the safety effect of the measure on target accidents.

The numerator of the cost-effectiveness ratio is estimated as follows: Number of accidents prevented (or expected to be prevented) by a measure = The number of accidents expected to occur per year X The safety effect of the measure

Page 242 SHORT TRAINING COURSE ON EFFICIENCY ASSESSMENT

The benefit cost ratio is defined as: Benefit-cost ratio = Present value of all benefits/ Present value of implementation costs When a CBA is applied, then, besides the above CEA’s components, the monetary values of the measure’s benefits are also required. The monetary values imply, first of all, accident costs and, depending on the range of other effects considered, may also include costs of travel time, vehicle operating costs, costs of air pollution, costs of traffic noise, etc. In order to make the costs and benefits comparable, a conversion of the values to a certain time reference is required. Such an action needs a definition of the economic frame, i.e. the duration of effect (length of service life of the project) and the interest rate, which are those commonly used for the performance of economic evaluations in the country. In a basic case, where the benefits come from the accidents saved only (and no influences on travel expenses and the environment are expected), the numerator of the benefit-cost ratio will be estimated as: Present value of benefits = Number of accidents prevented by the measure X Average accident cost X The accumulated discount factor, where the accumulated discount factor depends on the interest rate and the length of life of the measure. b. Safety effects The most common form of a safety effect is the percentage of accident reduction following the treatment. The main source of evidence on safety effects is from observational before-after studies. Other (theoretical) methods for quantifying safety effects are also possible. One should remember that the safety effect of a measure is stated as available if the estimates of both the average value and the confidence interval of the effect are known. One should also ascertain that both the type of measure and the type of sites (units) for which the estimates are available, correspond to those for which the CBA/CEA is performed.

Page 243 SHORT TRAINING COURSE ON EFFICIENCY ASSESSMENT

For WP4’s evaluations, it was desirable to apply the local values of safety effects, i.e. those attained by the evaluation studies performed in the country. When the local values do not exist, the summaries of international experience can be used26.

If the value of a safety effect is supposed to be provided by a current study (for which the CBA is performed), the estimation of safety effect should satisfy the criteria of correct safety evaluation. This implies that the evaluation should account for the selection bias and for the uncontrolled environment (e.g. changes in traffic volumes, general accident trends). c. Implementation units In the case of infrastructure measures, the appropriate unit will often be one junction or one kilometre of road. In the case of area-wide or more general measures, a suitable unit may be a typical area or a certain category of roads. In the case of vehicle safety measures, one vehicle will often be a suitable unit of implementation, or, in the case of legislation introducing a certain safety measure on vehicles, the percentage of vehicles equipped with this safety feature or complying with the requirement. For police enforcement, it may be a kilometre of road with a certain level of enforcement activity (e.g. the number of man-hours per kilometre of road per year); in the case of public information campaigns - the group of road users, which is supposed to be influenced by the campaign. d. Target accidents The accidents affected by a safety measure present a target accident group. Depending on the type of safety measure it can also be a target injury group, target driver population, etc. Target accidents depend on the nature of the safety measure considered. There are no strict rules for this case. For general measures like black-spot treatment, traffic calming, speed limits, etc. the target accident group usually includes all injury accidents.

26 Such as: Elvik R. and Vaa T (2004) The handbook of road safety measures. Elsevier. Page 244 SHORT TRAINING COURSE ON EFFICIENCY ASSESSMENT

One should remember that if we apply a specific and not general accident group, proper corrections should be performed for the accident costs, as well. e. Accident costs As known, a detailed survey of practice in estimating road accident costs in the EU and other countries was made by an international group of experts as part of the COST-research programme 27. Five major cost items of accident costs were identified as follows:

(1) Medical costs (2) Costs of lost productive capacity (lost output) (3) Valuation of lost quality of life (loss of welfare due to accidents) (4) Costs of property damage (5) Administrative costs

27 Alfaro, J-L.; Chapuis, M.; Fabre, F. (Eds): COST 313. Socioeconomic cost of road accidents. Report EUR 15464 EN. Brussels, Commission of the European Communities, 1994. Page 245 SHORT TRAINING COURSE ON EFFICIENCY ASSESSMENT

The relative shares of these five elements differ between fatalities and the various degrees of injuries, and also differ among countries. We assume that each country has its official valuations of accident injuries and damage. Otherwise, the comparative figures from the recent studies can be of help 28. All the values are applicable for the WP4’s evaluations but, in every case, there should be a clear indication which components of the above accident costs are included. For the sake of comparability of the evaluation results, the monetary values will be converted to € at 2002-prices.

The literature discusses mostly the valuations of fatalities and injuries whereas a CBA usually needs average accident costs. In a simple case, the average accident cost can be estimated as the sum of injury costs multiplied by the average number of injuries with different severity levels, which were observed in the target accidents’ group; the damage value per accident should be stated and added to the injury costs. f. Implementation costs The implementation costs should be determined for each safety measure considered. The implementation costs are the social costs of all means of production (labour and capital) that are employed to implement the measure. The implementation costs are generally estimated on an individual basis for each investment project. As no strict rules are available on the issue, performing a WP4’s evaluation, all the components of the implementation costs should be explained. Typical costs of engineering measures, which are recommended for the CBA evaluations in the country, are desirable.

28 see Chapter 2 of WP3’s Handbook Page 246 SHORT TRAINING COURSE ON EFFICIENCY ASSESSMENT

The implementation costs should be converted to their present values, which include both investment costs and the annual costs of operation and maintenance. Similar to the case of accidents costs, for the sake of comparability of the evaluation results, the monetary values will be converted to € at 2002-prices. g. Treatment of uncertainty In most cases, all effects, particularly the safety effects cannot be determined exactly. It is necessary to consider the level of uncertainty within the calculation, give exact figures and explain the variation of the results at their mean and at the borders of a (in most cases 95%) confidence interval. If uncertainties cannot be calculated or estimated, they have to mentioned at least and figures of the possible outcomes have to be described. h. Examples For a better understanding it is strongly recommended to use examples of well elaborated efficiency studies. It is also recommended, when using the short training course, to use other examples. Certainly, the examples used have to taken from well elaborated (according to the standards mentioned above, state-of-the-art EA) EA studies and be presented in a similar way as shown below for one of the ROSEBUD WP4 cases.

Page 247 SHORT TRAINING COURSE ON EFFICIENCY ASSESSMENT

Page 248 ROSEBUD WP4 - CONCLUSIONS

CONCLUSIONS by Victoria Gitelman and Shalom Hakkert Overview and Summary tables by George Yannis and Eleonora Papadimitriou WP4 descriptions by Martin Winkelbauer

1 Summaries of WP4 activities

The procedure adopted within ROSEBUD - WP4 was designed to gain experience from various countries in performing efficiency assessment (EA) studies of road safety related measures along the lines developed in earlier Work-packages. Particularly, the intentions were: • To test the availability of data and values for the performance of EA studies such as exposure data, accident data, etc; values of safety effects, accidents costs, implementation costs, environmental and other impacts. • To test the EA methods towards their applicability for road safety measures. • To perform EA studies of a considerable number of cases of safety-related measures ("case studies") which may serve as evaluation examples for similar cases, e.g. for the same or comparable road safety measures in other countries. • To examine the usability of procedures, methods and recommendations developed by the previous work-packages of ROSEBUD. • To gather problems which have not been targeted so far within the ROSEBUD framework and to develop solutions and recommendations. • To present the results of the case studies to decision makers, to document their feedbacks and to develop recommendations for such presentations and the assessment process and documentation as a whole. To cover all these goals, the following steps were undertaken: • Road safety measures were selected for assessment within WP4 (10 cases). • Among these measures, two cases were selected for detailed discussion with decision-makers. One of the cases was presented to a group of decision makers in a one-day workshop. The other case study was sent to a decision maker in a printed version. In both cases, the feedback of the decision makers was recorded and afterwards discussed within the workgroup. • In a one day conference (3rd ROSEBUD Conference, Vienna, March 18th, 2005), both cases were presented to a broader audience with a majority of the participants being members of the User Representation Group of ROSEBUD.

Page 249 ROSEBUD WP4 - CONCLUSIONS

1.1 The WP4 – workshop

On the 16th of December 2004, a workshop in Bordeaux hosted by CETE SO was conducted and dedicated to • receiving feedback on the "Short Training Course" for decision makers; • present case G "Measures Against Collision with Trees" to decision makers in order to get feedback on both the results of the study and the applicability of the efficiency assessment. • test if this approach can be applied to a larger audience, e.g. at the 3rd ROSEBUD Conference. The agenda of the Bordeaux-workshop was as follows: • Introduction of the decision makers and their role within decision making. • Description of ROSEBUD. • "Short Training Course" on efficiency assessment. • Presentation of the results of efficiency assessment (CBA) on measures against collisions with trees. • A broad discussion of the results focused on the usability of these results within the decision making process and the "Short Training Course". This was supported by a set of specific questions, which was developed specifically. The feedback from decision makers was recorded and discussed within the WP4 working group in a meeting on the next day. A procedure for the conference was developed and an agenda was drafted. Furthermore, the current status of all case studies was presented and discussed among the working group.

1.2 The 3rd ROSEBUD Conference

In accordance with the results of the WP4 – workshop in Bordeaux, the agenda for the 3rd ROSEBUD conference was prepared (excluding formal parts): • Intentions and Current Status of WP4 • A keynote lecture which addresses the need for integrating EA in the decision making process and encourage a fruitful discussion afterwards. • A "Short Training Course" on efficiency assessment. • Overview of all WP4 cases In two parallel sessions: • Results of the two case studies were presented and discussed • Presentation of the decision makers' impressions on those studies with specific respect to their feasibility within the decision making process. • A discussion of the case study results focusing on the usability of the results within the decision making process. In a plenary session: Page 250 ROSEBUD WP4 - CONCLUSIONS

• A plenary discussion including the two case studies and general issues of EA and decision making. Chairmen of the discussion were two decision makers and two experts. This was also supported by a set of specific questions, which was based on the questions used in the workshop and improved with respect of the feedback gathered there. • Preview on ROSEBUD - WP5 and the final results and products of ROSEBUD.

2 Discussion and Conclusions

2.1 An overview of the case-studies

Within the framework of ROSEBUD WP4, the efficiency of various road safety measures was assessed through case-studies conducted in different countries. The selected measures covered different road safety related categories, decision-making levels and target accident groups. The evaluations were in line with standard evaluation techniques, with additional adaptations if necessary. Table 1 summarizes the results of the EA analyses and the characteristics of evaluation methods applied. In total, within the WP4, 18 case-studies were carried out, which covered 10 groups of safety-related measures. Out of the 18 case-studies: - 3 cases concerned vehicle-related measures (fitting motorcycles with ABS; compulsory DRL for the whole year); - 9 cases concerned infrastructure-related measures (traffic calming measures in urban areas; grade separation of at-grade rail-road crossings; installation of roadside guardrails; introducing signal control at a rural junction; constructing 2+1 road sections) and - the remaining 6 cases concerned user-related measures (automatic speed enforcement; large-scale projects of intensive police enforcement; compulsory helmet wearing for cyclists). It can be seen that: • Enforcement-related measures appear to be more cost-effective than other measures, obviously due to lower implementation costs. The efficiency of other user-related measures and of vehicle-related measures is also relatively high due to the same reason (low implementation costs per unit of implementation). On the other hand, the efficiency of infrastructure-related measures varies widely, depending both on the construction costs and safety effects of the measures. • National-level measures are generally more cost-effective than local-level measures. However, this finding mostly stems from the fact that the majority of local-level measures are road infrastructure improvements. • No significant differences can be found in the efficiency of similar measures applied in different countries. • The target accident group/ target population usually includes all road accidents/ all drivers, with some obvious exceptions such as case A ("fitting motorcycles with ABS") for which "motorcycle riders" are the natural target population; case G ("implementation of roadside guardrails") which is dedicated to the prevention of roadside collisions with trees; case J ("2+1 roads") which struggles with head-on collisions; and case K which concerns bicycle riders only.

Page 251 ROSEBUD WP4 - CONCLUSIONS

• Typically, the accident costs come from official national data; in a few cases (mostly, Israeli and Greek case-studies on infrastructure-related measures and intensive police enforcement) some adaptations of the official injury costs were made to provide a valuation of an average accident. • The availability of implementation costs was problematic in many cases. Nevertheless, in the majority of cases the estimates of implementation costs were based on the official data provided by relevant authorities. In the cases where the evaluation was performed prior to the measure's implementation (e.g. ABS for motorcycles, DRL, compulsory helmets for cyclists) some practical assumptions or the valuations of similar measures applied in other countries (i.e. the "literature" source) were accounted for in the costs. • For the calculation of safety effects, before-after considerations with control-groups were the most common. In other cases, estimates from the literature or from previous research were applied. Only a few cases applied a number of simple assumptions, estimating the safety effect of the measure. • Additional (other than safety) effects were estimated in half of the cases. In some other cases a need to account for the additional effects was mentioned but not realized due to lacking data/ models which could isolate the effects (i.e. changes in air pollution, noise level, travel time or fuel consumption) associated with the measure.

Page 252 ROSEBUD WP4 - CONCLUSIONS

Table 80: Summary characteristics of the case-studies Level of Implemen Source of Category of implemen Target tation Accident safety effect measures tation group costs costs value Other effects CBA results

Nr. Benefits to Case Study Description of measure Local Noise costs ratio National Regional Literature Literature Literature All drivers Estimates Estimates Air pollution Official data Official data Official All accidents Assumptions Time savings Time Vehicle-related Fuel consumption Regression model User-related - others Infrastructure-related Before-after comparison Accident/ driver sub-group Accident/ User-related - Enforcement Country of measure Country Case-responsibility Fitting motorcycles with ABS and reducing ABS 1.1-1.4 A 1 ABS-Motorcycle √ √ AT AT taxes √ √√√ √√ 9.4-11.7 Automatic speed enforcement in a tunnel 5.4 B 1 Section Control √ √ AT AT (motorway) √ √ √ √ + 2 Section Control √ √ NL AT Automatic Speed Enforcement on a motorway √ n/a √ √ ++ n/a C 1 Daytime running lights √ √ CZ CZ DRL for the whole year √ √√√ √ - - 4.3 2 Daytime running lights √ √ AT CZ DRL for the whole year √ √ √ √ - - 3.6 E 1 Traffic calming (urban areas) √ √ IL IL Speed humps (1 road) √ √ √ √√ - 2.0-4.0 2 Traffic calming (urban areas) √ √ GR GR Speed humps, woonerfs (area) √ √ √ √√ - 1.14-1.2 3 Traffic calming (urban areas) √ √ CZ CZ Roundabouts instead of four-arm intersections √ √ √ √ 1.5 1.4 (urban) F 1 Rail-road crossings √ √ IL IL Grade separation of at-grade rail-road crossing √ √ √√ √√ ++ 2.8 (rural) 0.94 (urban) 2 Rail-road crossings √ √ FI FI Grade separation of at-grade rail-road crossing √ √√ √ ++ 2.5 (rural) Measures against collisions G 1 with trees (guardrails) √ √ FR FR Implementation of roadside guardrails √√ √ √ 8.7 Road improvement mix (rural H 1 areas) √ √ IL IL Introducing traffic signal control at a rural junction √ √ √√ √√ 1.25 5-year project (interurban roads), with emphasis on I 1 Intensive police enforcement √√ GR GR speed and alcohol √ √ √√ √ √√ 6.6-9.7 2 Intensive police enforcement √√ IL IL 1 year project (interurban roads) √ √ √√ √√√ 3.5-5.0 Constructing a 2+1 road section (without median √ J 1 2+1 roads √ √ FI FI cable) √√ √ 1.25 Constructing a 2+1 road section (with a median 2 2+1 roads √ √ SW FI cable) √√ √ √ + 2.26 Compulsory helmet regulation K 1 for cyclists √√ AT AT Compulsory bicycle helmet wearing √ √√ √ 1.14-2.28 Compulsory helmet regulation 2 for cyclists √√ DE AT Compulsory bicycle helmet wearing √ √√ √ 2.23-4.45

PAGE 253 ROSEBUD WP4 - CONCLUSIONS

2.2 The evaluation techniques applied

All the case-studies followed the standardised procedure of cost-benefit analysis (CBA). None of the studies selected the cost-effectiveness analysis (CEA) due to obvious limitations of the CEA when a single measure is evaluated and, especially, when the evaluation should also account for other (other than safety) effects. Besides, the discussions on the EA results with decision-makers seem easier when the results are presented in usual money-terms. None of the studies considered project alternatives; by default, each study compared "implementation of the measure" with a "do nothing" alternative. All other steps of the CBA evaluation procedure, i.e. a consideration of safety effects and side effects (on mobility and environment), monetising all effects, estimating implementation costs, calculation of present values of costs and benefits, and of efficiency measure (cost-benefit ratio - CBR) – were applied by the majority of the studies. The exceptions were basically due to lacking data. Estimating safety effects of the measures, the emphasis was put on the application of a correct safety evaluation. In the "ex-ante" evaluations the best available values of safety effects (which are based on a summary of previous experience/ research) were typically applied. In the "ex-post" evaluations, the safety effect value was typically estimated by means of the odds-ratio with a comparison group. A weighted value of the effect, based on the safety experience of a group of treated sites, was applied, when possible. In these cases, confidence intervals for the estimated safety effects were also provided. For the economic evaluation, typical scenarios adopted were "conservative" or "best estimate", although these were based on different approaches in each case. In some cases, different scenarios were dictated by several values of safety effects; in others – by a consideration of safety effects only versus a combination of safety effects with other side-effects. In any case, consideration of a number of scenarios appears to be useful for testing sensitivity of the results and, therefore, should be recommended for the usual evaluation practice. Summarizing the performance of the evaluation studies, several points can be mentioned indicating common technical problems which might occur during the CBA evaluations. They are: - a correct application of the odds-ratio technique, e.g. in the case of zero-values of some of the numbers; - ways for checking the statistical significance of the evaluation results; - the selection of side-effects to be considered along with safety effects; - a correct distinction between the implementation costs and negative side-effects of the measure (e.g. increased fuel consumption or travel time). For a more correct and uniform performance of CBA for safety-related measures it would be useful to elaborate a categorization of cases, indicating the types of impacts (e.g. safety, mobility, noise, air pollution) to be considered in the evaluation of each category of measures. For example, in the cases of infrastructure or enforcement measures, which have an implication on travel speeds, a consideration of changes in travel time would be useful. Another question concerns the inclusion of fines in the economic evaluation of

PAGE 254 ROSEBUD WP4 - CONCLUSIONS enforcement measures. A possible recommendation may be as follows: to fully include the investments made for enforcement measures in the costs is a necessary condition for consideration of fines as benefits. When a number of impacts are combined in the evaluation of a measure, a distinction should be made between the implementation costs and negative benefits of the measure. According to the recommended procedure (WP3, 2004), the implementation costs are the social costs of all means of production (labour and capital) that are employed to implement the measure, whereas the benefits include all effects which stem from the measure's application. Some benefits may be negative, e.g. increased travel time; in this case, their values are subtracted from the total benefits. Aiming at a better methodological basis of the evaluation studies as well as at a comparability of the results, it would be useful to address the above and other issues in the extended version of guidelines for the performance of the EA studies. In general, safety effects estimated should satisfy the criteria of a correct safety evaluation, i.e. to account for general accident trends, selection bias and possible confounding factors (e.g. changes in traffic volumes in "after" as opposed to "before" periods). The effect on accident numbers needs to be based on a comparison of the null hypotheses (accidents which would occur had no measure been taken) with actual accident numbers observed after applying the measure. A comprehensive theory of the topic is presented in Hauer (1997). The applicable techniques can be found in many publications (e.g. Elvik, 1997; Elvik, 1999). It is believed that a distribution of a brief guide on standardized techniques for the evaluation of safety effects would be helpful for safety practitioners, in general, and particularly, for the improvement of quality of the EA studies.

2.3 The EA components: data and values

Generally, accident data were easily accessible to the authors of the EA studies. The valuations of road accident injury costs are usually provided by recently published evaluation studies. However, it was more difficult to attain costs of road safety measures. In the cases of infrastructure improvements and enforcement projects, the investments are paid from public budgets, therefore it frequently appears difficult to determine total values of these costs. Consultations with the responsible decision-makers and/ or analysis of valuations from similar studies may serve as the sources of values in this case. Establishing databases with typical implementation costs of safety improvements seems to be a practical solution for the systematic use of these values for EA studies. While the "ex-post" studies typically estimate the actual safety effect which can be associated with the application of safety measures, the "ex-ante" studies apply the available values, which should be based on previous research. To stimulate the application of more uniform and well-based values of safety effects, it would be useful to establish a database with typical values of the effects, based on international experience. Such a database might be open to a European network of experts and provide for general values of safety effects on initial steps of CBA/CEA as well as assist in judging the local effects observed. Lack of models for evaluating side-effects associated with the safety measure (i.e. changes in air pollution, noise level, travel time or fuel consumption) and, sometimes, lack of local valuations of theses effects, deter the consideration of theses effects by the EA studies. The problem may be tackled by a systematic accumulation of recommended

PAGE 255 ROSEBUD WP4 - CONCLUSIONS values and solutions (depending on safety measures considered) within the guidelines for the EA performance.

2.4 Role of barriers

The fundamental (or absolute) barriers to the application of the EA to road safety measures were left beyond the scope of the current consideration. None of the decision- makers involved rejected the principles of efficiency assessment. Concerning the local level of decision-making some experts doubted the practical influence of the evaluation results, however, not because of a principle non-acceptance of the approach but mostly due to the awareness of other factors (political, emotional) which usually influence such decisions. On the other hand, the relative barriers (of institutional or technical nature) did influence the cases' performance. The technical barriers such as typical problems with the evaluation techniques or lacking data (as mentioned above) were generally overcome by the evaluation studies. In some cases, thoroughly based statistical models were developed to ascertain the lacking values of the effects. In general, the majority of technical barriers, which might appear during the performance of an EA study, seem treatable. A lack of obligatory procedure for the performance of cost-benefit evaluations of safety effects is known as a major institutional barrier for the application of the EA of safety measures. However, in many cases (mostly, "ex-post" evaluations of enforcement and infrastructure measures) the CBA results emphasized the accident reduction effects and the economic savings associated with the measures' application. As a result, the decision- makers were interested in the distribution of the EA results and in further performance of the analyses. As to the barriers for implementation of safety measures, which were evaluated by the studies and found effective in the majority of cases, different forms of these barriers were identified by the studies. The wide application of the measure is frequently limited due to economic reasons (lack of finance, high costs, etc). Sometimes, safety reasons may conflict with other considerations (e.g. environmental issues like in case G – “measures against collisions with trees”). In other cases (e.g. helmets for bicycles, DRL, automatic speed enforcement) lack of publicity support or lack of acceptance by the general public deters the decision-makers from the measure’s promotion. However, in several cases (e.g. DRL for the Czech Republic, grade-separation of rail-road crossings in Israel, traffic calming in urban areas in Greece) the CBA results highlighted the expected/ attained benefits of the measures and, in this way, contributed to the acceptance of the measure by the decision-makers.

2.5 The usefulness of efficiency assessment for decision-making

Frequently, consideration of EA is part of the preparation of regional or local road safety plans. At the initial stage of evaluation, safety effects are usually unknown. To influence any decision making process, EA studies have to be prepared ex-ante using impact data from similar other measures taken from somewhere else. This stresses the need for availability and accessibility of evaluation studies on road safety measures as well as dissemination of EA results on an international basis. Authors of efficiency studies should be encouraged to use results from similar cases for this purpose.

PAGE 256 ROSEBUD WP4 - CONCLUSIONS

In some cases, safety studies of road infrastructure measures are required to justify a choice among different solutions to the same problem. EA can be very useful for decision making in such cases, including the taking into account of other, non-safety, effects and costs. At the local level, the application of a safety measure is in many cases not just an economic question but also a matter of subjective judgement. This problem can occur where the program of "good measures" is developed at the national level but executed at regional or local level. Benefits estimated at the national level are frequently not visible at the local level, where costs and local political interests dominate the decision makers' perspective. During the preparation of EA studies within such an environment, the financial benefits need to be explained considering the level of future decision making in the best possible manner. As stated by one local decision-maker on the local level, not the millions of Euro expected to be saved, influence the decision but the fact that somebody familiar to the decision maker was killed in an accident. This highlights the conflict between traditional arguments used in decision making and EA as an instrument to be promoted. As mentioned above, decisions at the local level involve a mix of global and local interests. In presenting the study results it is important to fit the arguments to the level of decision- makers. This comment refers to the specific situation of national road safety programs applied at regional or local level. To preserve the intentions of the national safety programs, the arguments need to include a presentation which is useful for the promotion of the original intentions at the regional or local level. The difference in usefulness of CBA versus CEA will also very much depend on the formal process of funding. As far as the French model, which was discussed at the workshop in Bordeaux, was concerned, the question of selecting guard-rail installation versus tree felling could be based on CEA, but again, emotional arguments were dominating the negotiations in the detailed planning process. Local decision makers in charge of road safety decisions seem to think that issues other than casualties (i.e. mobility costs, time use, environmental costs) will hardly be of use in local decision-making. In general, the feelings resulting from the discussions with local safety decision makers is that EA should be more directed to road safety and economic experts than to local decision makers. In the countries where the safety budget is centralized (i.e. the majority of local safety projects are financed by the government), the requirement of a CBA of safety measures may be distributed by stating it as a necessary condition for the application of projects coming from the central budget. CEA can be more applicable at the local level as no comparison with conflicting targets is usually performed and needed. The method of CBA at lower levels of decision making appears to be quite abstract. Specifically, in discussion with e.g. local peer groups, benefits at the national or even global level are weighted low or even disregarded, since impacts are not visible at the local level. The WP4 workshop showed the importance of decision markers' understanding of the principles of EA. The "short training course" was helpful on this issue. Some decision-makers voiced the opinion that when politicians make decisions they do not want to have too much input for these decisions. Elaborate EA studies narrow their range of decisions. Therefore EA seems to be "actively disregarded" or even objected to in

PAGE 257 ROSEBUD WP4 - CONCLUSIONS general. Rather cynically another opinion stated that EA is welcome as long as the results support the intentions of decision makers.

2.6 The form of presentation of case study results

On the basis of discussions with the decision-makers it was found to be useful to elaborate the presentation forms – the summary forms with the EA results, for different levels of decision makers. For laymen and non-professionals, the presentation of case results should be rather short. Figures on fatalities usually have a strong effect on decision makers. It is recommended to present local decision-makers with just one sheet (one page-presentation) of data, which should include a comparison of before and after accidents. The whole case report is needed when dealing with topics of national concern. At a higher level of decision-making the information presented needs to be more detailed. More detailed information improves the quality of the background material and improves the quality of decision-making. Presenting (marketing) the results, it is important to make a distinction between "technicians" (the professional level) and others. The language should be adapted to the targeted population. The educational background and function of the recipient need to be considered. For the professional level of decision makers it is important to explain the framework of components, which should be performed depending on categories of safety measures evaluated. From the various contacts with experts and decision makers in WP4, different suggestions were made concerning the amount of information that should be presented as a result of an EA study. It was felt that only in a small share of the cases, presenting all results of a study will be the optimum. Some voices recommended preparing a one-page information sheet. Frequently, the working group members received suggestions, only to present a rating of road safety measures (comparable to "star-ratings" e.g. used for safety of new cars), which would be very striking, particularly in discussions at local level and with the public. In summary: each recipient needs to be treated with an individual presentation of the results and individual background information adapted to the recipient. Although it was frequently stated, that the higher the level of decision making, the stronger the need for comprehensive information, the other issues mentioned above also have to be considered in each unique case. An important question is how to present the results to the public. In general, it can not be supposed that the public understands all the methods and processes of EA. Therefore, results need to be simplified to forward an understandable message to the public. While economic valuation of injuries and (particularly) fatalities may be accepted among experts, the average citizen is likely to oppose a monetary valuation of life. The presentation of EA results to the public needs to be carried out very carefully to avoid public resistance against the basic principles of EA.

2.7 Distribution of knowledge

The WP4 workshop again showed the importance of decision markers' understanding of the principles of EA. Training is also needed for those carrying out EA studies. There is a

PAGE 258 ROSEBUD WP4 - CONCLUSIONS need for international standards (guidelines) for preparing such studies. Both shall improve the quality of EA studies. Within the frame of the ROSEBUD Thematic Network such guidelines will be prepared. Deriving from WP4 experience, experts should be encouraged to publish their evaluation results on effects of road safety measures and results of EA studies. Reports should be inserted in international library databases (e.g. the ITRD) to become internationally available. To enable information exchange in day-to-day-business an internet forum for EA related issue could be installed. Possible ways for the dissemination of ROSEBUD results and messages in a country may be in the form of a workshop for national decision-makers, which includes: (a) a training course on principles of the EA of road safety measures; (b) the results of evaluation studies performed for local conditions. One of the most important findings within the practical testing done in WP4 is that the presentation of EA results has to be set up in close relation with the recipients. The level of decision making (international, national, regional or local), function (experts, researchers, government employees, politicians, etc.), educational background (lawyers, engineers, economists, etc.) and even individual characteristics of the recipient need to be considered. Particularly, it is recommended to take in account their personal experience and knowledge in the field of EA. The need for preparing the presentation of EA studies is very diverse, ranging from full training on EAT to no information at all.

2.8 Recommendations

Recommendations addressing the “best practice” guidelines and the evaluation framework in general: • Further development of the EA procedures and methods is required. • Particularly, for a more correct and uniform performance of CBA for safety-related measures it would be useful to elaborate a categorization of cases, indicating the types of impacts (e.g. safety, mobility, noise, air pollution) to be considered in the evaluation of each category of measures. • Safety effects estimated should satisfy the criteria of correct safety evaluation. A distribution of a brief guide on standardized techniques for the evaluation of safety effects would be helpful for safety practitioners, in general, and particularly, for the improvement of quality of the EA studies. • The implementation costs of safety measures are usually lacking. Establishing databases with typical implementation costs of safety improvements would be of help for the systematic use of these values in the EA studies. • A database with typical values of safety effects, based on international experience would be useful for correct and systematic performance of the "ex-ante" studies. • Consideration of a number of scenarios is useful for testing sensitivity of the results and should become common for the usual evaluation practice. • Definition and main components of a mini-CBA as well as its applicability for different levels of decision-making should be clarified. • It is important to clarify the definitions of projects for which the EA of safety impact should be performed. It is suggested that the EA of safety impacts should be

PAGE 259 ROSEBUD WP4 - CONCLUSIONS

applied mostly for two types of projects: (a) the improvements which were financed by safety-dedicated budgets and (b) the projects aimed at improving safety.

Recommendations addressing the distribution of EA procedures/ evaluation results: • It would be useful to elaborate the presentation forms – the summary forms with the EA results, for different levels of decision-makers. • Presenting the results, it is important to make a distinction between "technicians" (the professional level) and others. The language and the details should be adapted to the targeted population. • CBA seems to be more suitable for national- and regional-level decision-making where the safety budgets are planned. CEA seems more suitable for local level, especially when several safety solutions are compared while tackling a specific safety problem. • In the countries where the safety budget is centralized, an EA of safety measures may be distributed by stating it as a necessary condition for the application to central budget. • Training of decision-markers is important to strengthen their understanding of the principles of EA. Training is also needed for those carrying out EA studies.

References Elvik, R. (1997). Effects on Accidents of Automatic Speed Enforcement in Norway. Transportation Research Record 1595, TRB, Washington, D. C., pp.14-19. Elvik, R. (1999) Cost-benefit analysis of safety measures for vulnerable and inexperienced road users, Work package 5 of EU-Project PROMISING, TØI-Report 435, Institute of Transport Economics, Oslo. Hauer, E. (1997). Observational Before-After Studies in Road Safety. Pergamon. WP3 (2004) Improvements in efficiency assessment tools. ROSEBUD.

PAGE 260 ANNEXES

ANNEXES

Annex 1:Set of Questions used at the WP4 Workshop • Short statement of the decision-makers: What are your opinions on the case? • Our comments (of the team)? • "Interesting questions" - opinions. • What questions do you expect, if you take the results of this CBA to another forum? • What do you expect when these results are published by mass media? • What do you expect from a public discussion in general? • Is there any information missing for the further decision-making process? • Which will be the most critical points in further discussion - among decision-makers, among experts and in the public.

Annex 2: Set of Questions used for the panel discussion during the 3rd ROSEBUD Conference General Topic: • How useful is EA for decision-making? Concerning the method: • Can we exchange value of life for time savings? Concerning dissemination to different audiences: • Would you use EA and the results based on EA in your own communication with third parties? • How will the public accept EA? • Is EA understood as an objective instrument? • Which will be the most critical points in further discussion of EA results? Concerning the usability in practical decision making: • What is needed to make EA practically useful? • How will EA influence the decision-making process? • Can decision-makers be convinced by EA? • Will the use of EA tools improve road safety efforts?

PAGE 261