JUNE 2021 1 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS MOBILITY AND TRANSPORT CONNECTIVITY SERIES DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? Lessons from the Republic of Korea’s Recent Move to Lower Speed Limit on Urban Roads 2 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? © 2021 The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissem- ination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Disclaimer—This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they repre- sent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. All queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202- 522-2625; e-mail: pubrights@worldbank.org. Cover photo credits: World Bank DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? Lessons from the Republic of Korea’s Recent Move to Lower Speed Limit on Urban Roads Sudeshna Mitra and Soames Job, Global Road Safety Facility, World Bank Sangjin Han and Kijong Eom, Korea Transport Institute Contents 1. Introduction and Background..............................................................................................................................................6 2. Objectives and Scope..............................................................................................................................................................9 3. Literature Review.....................................................................................................................................................................11 Speed and Safety Risk......................................................................................................................................... 12 Methods of Evaluation of Effectiveness........................................................................................................... 15 4. Methodology and Data Sources........................................................................................................................................18 5. Results........................................................................................................................................................................................ 24 Assessment of Safety Benefit from Speed Limit Reductions......................................................................... 25 Before vs. after comparison of total crashes................................................................................................................... 25 Before vs. after comparison of vehicle-to-vehicle crashes............................................................................................. 27 Before vs. after comparison of vehicle-to-pedestrian crashes...................................................................................... 28 Assessment on Impact of Travel Time from Speed Limit Reductions.......................................................... 30 6. Conclusions and Policy Recommendations................................................................................................................31 Figures Figure 3.1. How a Change in Impact Speed Changes the Risk of Fatality, by Different Crash Types...................................13 Figure 4.1. 30 kph Limit in Residential and Commercial Neighborhoods (with Pavement Marking)...................................21 Figure 4.2. Pedestrian Priority Zone with 20 kph Limit and Modified Pavement Texture.....................................................21 Tables Table 4.1. Current Status of Lowering Speed Limits across Korea............................................................................................22 Table 4.2. The Current Status of Lowering Speed Limits in Daegu Metropolitan City...........................................................23 Table 5.1. Before vs. After Comparison of Total Crashes...........................................................................................................25 Table 5.2. Calculation Table for Comparison of Crashes Using the Before-After Method with Control Groups.................26 Table 5.3. Before vs. After Comparison of Vehicle-to-Vehicle Crashes.....................................................................................28 Table 5.4. Before vs. After Comparison of Vehicle-to-Pedestrian Crashes ..............................................................................29 Table 5.5. A Comparison of Travel Time across Study Sections in Daegu Metropolitan City.................................................30 6 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? 1. Introduction and Background 7 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS In many countries, speeding has become one of the In countries experiencing high motorization and most important risk factors contributing to road development, the lack of regulations results in sub- traffic fatalities and injuries. When vehicles are driven optimal speed limits for safety and mobility. For at high speed, the likelihood and severity of crashes instance, according to data from the World Bank’s increases, which leads to speed management becom- recently published “Guide for Road Safety Opportuni- ing an essential tool for improving road safety (ITF ties and Challenges: Low- and Middle-Income Coun- 2018). An important component of speed manage- try Profiles” (Wambulwa and Job 2020), currently very ment is setting the right speed limits, as they convey high speed limits are posted on urban roads in most the maximum legally permissible travel speed to the low and middle-income countries (LMICs), where road user, given the type of road, set up, and the type interactions with vulnerable road users are signifi- of road users who most often use the facility. As one cant. In such environments, the recommended Safe of the most fundamental rules of road traffic opera- System speed is no more than 30 kilometers per hour tions, speeds greatly influence not only traffic safety (kph), but no low-income countries (LICs), and only and operations, but also climate impacts and air and 3 percent of middle-income countries (MICs), have noise pollution (Sakashita and Job 2016). By reduc- posted speed limits of 30 kph or less on urban roads. ing the likelihood of vehicle-to-vehicle conflicts and Yet, most of the available studies on speed regulation conflicts among vehicles, cyclists, and pedestrians, and management have been conducted in higher an appropriate speed limit can enhance the safety of income countries such as Australia, the United States, driving as well as walking, cycling, and motorbike rid- and western Europe. The Republic of Korea (herein- ing. Indeed, setting safe speed limits, along with the after “Korea”), which regulates speed limits through supporting infrastructure and enforcement, is one the Road Traffic Act (specifically, Article 19 of the of the most important ways to enhance traffic safety Enforcement Regulations), provides an example of and secure the efficiency of mobility on roadways. an Asian country seeking to change and improve its road usage regulations. Many countries have established speed-limit regu- lations based on roadway classifications, observed In April 2016, the National Police Agency in Korea traffic patterns, and the intended usage of the road established the Transportation Infrastructure Con- (or, more appropriately, actual use of the road and struction Basic Plan, in which for the first time, the surrounding land use). Yet regulatory practice varies Safe Speed 5030 policy was included for urban greatly from country to country, even with compelling pedestrian safety. For its effective implementation, evidence-based support for specific optimal speed lim- subsequently, the 5030 Council was formed at the its on a very wide range of road geometries and the pan-governmental level, comprising the Ministry surrounding conditions, such as land use, the pres- of Land, Infrastructure, and Transport (MOLIT) and ence of roadside objects, and users such as pedes- relevant agencies. In the same year, the 8th National trians, cyclists, and motorcyclists. For example, most Road Safety Basic Plan was presented, acknowledg- developing countries—including many of the World ing the “expanded application of urban speed limit Bank’s client countries—lack regulations or even the 50/30” as an intermediate implementation measure, criteria for determining appropriate speed limits. emphasizing the need for speed management for the achievement of the goals of transportation safety. 8 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? This was entailed, in April 2019, by an amendment also experiencing accelerated development and to the Approved Code of Practice of the Road Traf- increased travel demands in urban areas. fic Act, adjusting the maximum speed to 50 kph as the standard speed limit on roads in urban areas, With this in mind, the Korea Transport Institute which is awaiting its official enforcement in 2021. In (KOTI) and the World Bank conducted a joint recent years the country’s National Policy Agency has research project to assess the safety benefit changed the speed limit of many urban roadways to of lowered speed limits on urban roads and to 50 kph or 30 kph to enhance traffic safety and main- inform future policy development on speed limits tain smooth driving. in urban areas. Specifically, this study analyzes how changes in the speed limit affect safety performance As a result of the speed limit control policy, the World and operational performance. Because speed lim- Bank and other international organizations have its influence the operating speed of the facility and focused on improved urban road safety in Korea, and affect both safety and operational efficiencies, the hope to transfer the experience of the speed policy study team hypothesized that studying the effect of and associated best practices to other developing changes in speed limit might yield valuable demon- countries through joint research. Furthermore, les- stration examples in the context of Asia and other sons from an Asian country—namely, Korea—could developing countries. be more acceptable to neighboring Asian countries References ITF (International Transport Forum). 2018. Speed and Crash Risk. Paris: Organisation for Economic Co-operation and Development (OECD) Publishing. https://www.itf-oecd.org/sites/default/files/docs/speed-crash-risk.pdf. Sakashita C. and R. F. S. Job. 2016. “Addressing Key Global Agendas of Road Safety and Climate Change: Synergies and Conflicts.” Journal of the Australasian College of Road Safety 27 (3): 62–8. https://acrs.org.au/wp-content/uploads/ Contributed-Articles-Addressing-key-global-agendas-of-road-safety-and-climate-change-synergies-and-conflicts.pdf. Wambulwa, William Majani and Raymond Franklin Soames Job. 2020. “Guide for Road Safety Opportunities and Challenges: Low- and Middle-Income Country Profiles.” Working Paper, World Bank, Washington, DC. http://documents.worldbank.org/curated/en/447031581489115544/ Guide-for-Road-Safety-Opportunities-and-Challenges-Low-and-Middle-Income-Country-Profiles. 9 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS 2. Objectives and Scope 10 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? Significant research has been undertaken on how 1. Evaluate the effectiveness of the reduced speed changes in speed limit—for example, the introduction limits in terms of crash reduction through a of 30 kilometers per hour, or kph (20 miles per hour, before-after study. or mph) speed limits—impact safety both when com- bined with, and without “traffic-calming” engineering 2. Examine if the speed limit change had different treatments such as speed humps or raised platforms. effects across different crash types, user types, However, most of the studies have been conducted in and crash severities. Australia or countries in Western Europe, with almost no recorded studies from Asia, Africa, the Americas 3. Evaluate the impact of speed limit change on and Eastern Europe. Though it may be reasonable, a transit speed through a before-after assessment. well-developed infrastructural environment such as that found in Korea would expect similar results as 4. Develop appropriate and actionable recommen- that of the western countries, a study originating in dations for departments of transportation in Asia could have a strong demonstration effect and developing countries. prove very convincing for many Asian countries. To start, this report first provides a brief literature With this in mind, the main aim of this study1 is to review on the concept of Safe System speed limits, present the findings from Korea’s reduced speed and the effect of speed limit reductions as part of limits on safety performance and to support estab- speed management in several countries, followed by lishment of suitable speed-management strategies a brief description of the evaluation methods for the based on a quantitative data-driven approach. The before-after assessment. This is followed by a sum- scope of the project was as follows: mary of the findings, a set of recommendations, lim- itations of this study, and finally, a capsule of future research that could be undertaken to either extend or follow up on the study. 1 The study was funded by the Korea Transport Institute (KOTI). World Bank staff involvement was funded by the World Bank and the Global Road Safety Facility (GRSF), with UK Aid funding. 11 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS 3. Literature Review 12 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? Speed and Safety Risk Speeding is one of the most significant risk factors for crashes possible at urban intersections, where pedes- road traffic fatalities and injuries all over the world. trians and other vulnerable road users are safely Because driving vehicles at higher speeds increases separated from traffic. Likewise, head-on crashes both the likelihood and severity of crashes, speed predominate on rural two-lane undivided highways management is recognized as an essential tool for (ITF 2018). The graph in figure 3.1 shows how the risk improving road safety (ITF 2018). of a fatality rises with an increase in impact speed, for four different crash types: pedestrian crashes, Out of several different speed management meth- crashes into rigid objects, side-impact crashes, and ods, lowering the posted speed limit using a static head-on crashes. sign (and enforcing that lower posted speed limit) is one of the more common, and effective, road-safety interventions. As a widely adopted speed manage- ment measure, static signage is often associated with or without other engineering measures aimed at producing lower vehicle speeds and crash and injury severity reductions. In addition, speed management is central to the Safe System approach of road design, which views human life and health as more important than anything else. Thus, the consequences for speed management of adopting a Safe System approach result in certain maximum speed limits, which depend on the road environment and the most predominant risk in such environment. For example, a speed limit in urban built-up areas with a mix of vulnerable road users and vehicular traffic will be very different than the limit posted on urban roads with signalized intersec- tions and limited interactions with vulnerable road users—and will vary even further from the limit on rural two-lane, undivided roads. In urban environments in low and middle-income countries (LMICs), interactions between pedestrian and other vulnerable road users and motorized traf- fic is often very high, mainly in the absence of infra- structure to separate them from high speed traffic. On the other hand, side-impact crashes, are the most frequent and most severe kind of vehicle to vehicle 13 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS Figure 3.1. How a Change in Impact Speed Changes the Risk of Fatality, by Different Crash Types Source: NSW Centre for Road Safety (n.d). Figure 3.1 clearly illustrates the dramatic effect of • 40 kph for impacts with solid objects speed on the risk of fatality in virtually any type of crash situation. The results strongly mandate safe • 50 kph for car-on-car side-impact crashes speed limits for different road environments, which are commonly known as Safe System speed limits. On • 70 kph for car-on-car head-on crashes the graph, the “Safe System speeds” are set at a point that enables a 90 percent survival rate (or 10 percent Even these limits may be considered too high to fatality rate): achieve a 90 percent survival rate, not even consid- ering the inevitably much higher rates of serious • 30 kilometers per hour (kph) for impacts with injury, more so for vulnerable road users who have pedestrians (and other vulnerable road users no physical protection. Since creating low-speed such as cyclists) environments to provide a greater degree of safety for vulnerable road users is one of the major aims 14 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? of the speed management program, physical speed Layfield 2003), suggest implementation of 30 kph (20 restraint measures, along with infrastructure and mph for the study) zones with physical traffic calming engineering measures, are often added to achieve measures could result in substantial reductions in higher levels of road user compliance in addition to average speed of around 14 kph, or 9 miles per hour posting lower speed limits. With this treatment, the (mph), and sizeable reductions in collisions. In both reduction in serious injuries to pedestrians can be studies, speeds before these schemes were intro- very high—in excess of 70 percent (Woolley et al. duced averaged around 40 kph (25 mph), dropping to 2018; FHWA 2020)—with significant benefits for road well below 30 kph (20 mph) after implementation. users as well. Comparison of before-after crash data from the first Crucially, the safety benefits of a speed limit reduc- study, for a total of 72 schemes, showed average tion depend very much on the magnitude of the annual crashes fell by 60 percent, while child pedes- change and the level of compliance produced by trian and cyclist crashes fell by 70 percent and 48 both engineering and enforcement mechanisms. percent respectively. In the second study, conducted Speed limits exert a major influence on actual driv- by Webster and Layfield (2003), which looked at crash ing speeds. However, the changes in speed observed data for 78 schemes, a before-after comparison indi- when introducing a new speed limit will rarely be cated a lower frequency of injury crashes within the strictly proportional to the change in the speed limit, zones by about 42 percent, and a lower frequency of as presented by Elvik (2012) based on a meta-analy- fatal or serious injury (killed or seriously injured, or sis. Elvik concluded speed almost always changes in KSI) crashes by about 53 percent. However, results the same direction as speed limits, but in almost all from a 2018 study evaluating the effectiveness of 20 cases, a reduction in the speed limit of, for example, mph (sign-only) speed limits reductions (Atkins Ltd., 20 kph, would result in an approximate 8 kph change AECOM, and Mather 2018) in the United Kingdom in mean traffic speed. A 10 kph reduction in the was somewhat inconclusive. The study findings indi- speed limit results in a much smaller change in actual cated that except in one case study location (Brigh- traffic speeds, typically around 2.5 kph. However, the ton Phase 1) evidence showed no significant short- good news is that a 10 kph reduction in the speed term change in collisions and casualties. The results limit could be expected to produce about a 15 per- from the Brighton Phase 1 study, however, showed cent reduction in injury crashes, and upwards of a 40 promise where a blanket 20mph limit was introduced percent reduction in pedestrian fatalities and serious covering both major and minor roads. In addition, injuries; however, evidence shows the benefits can be data also indicated a statistically significant change even greater in the right circumstances (McTiernan in collisions and casualties relative to the 30 mph et al. 2015; Turner, Howard, and Breen 2015; Jurewicz comparison group, concluding a significant reduction 2010; Elvik et al. 2009). in overall collisions (–18 percent), overall casualties (–19 percent), pedestrian casualties (–29 percent), and For example, two extensive studies undertaken by casualties aged 75 or over (–51 percent). The study UK-based TRL, formerly the Transport Research suggested the need for further data collection to Laboratory (Webster and Mackie 1996; Webster and determine the longer-term impact of 20 mph limits. 15 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS Another recent study of speed limit reduction pro- Road traffic operational environments in LMICs, how- gram from the Latin American city of São Paulo, Brazil ever, are very different than in high-income countries (Ang, Christensen, and Vieira 2020) also concluded (HICs), with a mix of vehicles and other road users the program of regulatory change in speed limit sharing road space—fleets of cars along with bicy- resulted in 1,889 averted crashes within the first 18 cles, trucks, and other modes of transport, in addi- months of such change, reducing crashes by 21.7 tion to pedestrians. Controlling or managing vehicle percent on treated roads, with larger effects on roads speeds is a complex and challenging task because of with camera-based enforcement. factors such as inadequate enforcement, a different road safety culture, and/or lack of driver discipline. While speed limit reductions show enhanced safety outcomes, additional studies demonstrate the harm While most of the studies mentioned earlier (except caused by speed limit increases. A study conducted the Brazil study) were undertaken in the context of by Farmer (2017) to examine the safety effects of HICs, speed management follows the basic rule of increases in U. S. state maximum speed limits during physics and should be applicable everywhere. The lit- the period from 1993 to 2013 associated a 5 mph (8 erature reviews also did not find any such speed limit kph) increase in the maximum state speed limit with reduction programs in the Asian context. As a result, an 8 percent increase in fatality rates on interstates experience and lessons from an Asian country in and freeways, and a 4 percent increase on other this context will add immense value—not only to the roads. In total, an estimated 33,000 more traffic fatal- existing state of knowledge, but also for speed man- ities occurred than would have been expected with agement in World Bank client countries that have a the lower speed limits. similar or at least comparable context. Another study from Victoria, Australia (Sliogeris and Roads Corporation 1992), assessed the effects of an increased speed limit (from 100 kph to 110 kph) on rural and outer Melbourne freeway networks in 1987, Methods of Evaluation which was later reduced in 1989. Results indicated an increase in injury crash rate (including fatalities) of Effectiveness per kilometer traveled of 24.6 percent when the Because monitoring and evaluation of road safety speed limit was raised in 1987, and a decrease of 19.3 interventions—or “treatments”—is as important percent when the limit was lowered in 1989. These as the selection of the interventions, to assess the results were in comparison to a control group. Finally, effectiveness of a treatment researchers rely on one more informative piece of evidence (Graham and three primary evaluation methods or approaches, Sparkes 2010), from New South Wales, Australia, indi- each ranging in complexity: 1) cross-sectional stud- cated a 45 percent reduction in vehicle-and-pedes- ies 2) before-after studies, and 3) experimental trian crashes, after implementation of an initiative to before-after studies. reduce speeds to 40 kph in school zones. 16 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? Cross-sectional studies compare safety perfor- the comparison sites are subject to the same sets of mance of sites with a particular safety feature (or external variables, any difference in safety outcomes features) with sites that do not have the same fea- must be due to the treatment. ture. These studies assume the difference in safety performance is due to this feature (or lack thereof). In road safety literature, however, the most recom- Using the cross-sectional approach can be problem- mended approach for evaluation of infrastructure atic, as it fails to control external factors, and does interventions is the Empirical Bayes (or EB) method not adequately handle regression to the mean. Thus, as explained by Hauer (1997). This method has been the cross-sectional approach is not recommended for designed to account for regression to the mean and evaluation of treatments. other factors that could affect crash outcomes with- out random allocation of the sites to treatment and Before-after studies may or may not involve compari- nontreatment conditions. The procedure is based on son sites. The most basic form of evaluation is to sim- the premise that what would have happened with- ply compare crashes in the period before an interven- out a treatment can be estimated by extrapolating tion is installed with the crashes after (termed as a what happened at the trial sites in the past, adjusted simple before-after or naïve before-after study—that to make allowance as far as possible for extraneous is, without a comparison group). This approach is also factors such as traffic flow, among others, and more not recommended, as it does not adequately account general changes affecting the entire road network for regression to the mean or external variables. (for example, weather, compliance with road rules, or changes to vehicles). Safety performance functions Before-after studies with comparison groups, is are also developed to estimate the expected number the most commonly applied method for the eval- of crashes for a site. However, this method requires uation of infrastructure treatment. , comparison more data, which is often a practical challenge. groups. Even though this approach does not fully address the issue of regression to the mean (how- Experimental before-after studies offer the most ever, using a longer “before” time period can reduce control, and allow investigators an active role in this effect), it does limit the impact of external factors. influencing the selection of sites for treatment. This The approach compares the outcomes at the treat- method usually takes the form of randomly allocating ment group with the outcomes at a set of compari- sites to the treatment group or the control group, the son group (sometimes termed “control” sites), which “gold standard” in evaluation methodology. However, have similar characteristics to the treatment sites in such an approach is very rare in evaluation of infra- all important aspects, except that the treatment is structure interventions, as many external elements not installed. This approach assumes external factors are beyond the control of the investigator in most act on both the treatment sites and the comparison observational studies, not to mention the ethical sites in an identical way, and so each can be allowed aspect that high risk sites cannot be left untreated.1 for and measured. Since the treatment sites and 1 For more details on evaluation methods, please refer to the Association of Australian and New Zealand Road Transport and Traffic Authorities (Austroads) report, An Introductory Guide for Evaluating Effectiveness of Road Safety Treatments (Cairney, Turner, and Steinmetz 2012). 17 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS References Atkins, Ltd., AECOM, and Mike Maher. “20mph Research Study: Process and Impact Evaluation Headline Report 2018.” Prepared for the Department for Transport, United Kingdom. https://assets.publishing.service.gov.uk/government/ uploads/system/uploads/attachment_data/file/757307/20mph-headline-report.pdf. Ang, Amanda, Peter Christensen, and Renato Vieira. 2020. “Should Congested Cities Reduce Their Speed Limits? Evidence from São Paulo, Brazil.” Journal of Public Economics 184 (April): 104155. https://doi.org/10.1016/j.jpubeco.2020.104155. Cairney, Peter, Blair Turner, and Lisa Steinmetz. 2012. An Introductory Guide for Evaluating Effectiveness of Road Safety Treatments. Publication No. AP-R421-12. Sydney, Australia: Association of Australian and New Zealand Road Transport and Traffic Authorities (Austroads). https://austroads.com.au/publications/road-safety/ap-r421-12. Elvik, Rune. 2012. “Speed Limits, Enforcement, and Health Consequences.” Annual Review of Public Health 33: 225–38. https://doi.org/10.1146/annurev-publhealth-031811-124634. Elvik, Rune, Alena Høye, Truls Vaa, and Michael Sørensen. 2009. The Handbook of Road Safety Measures 2nd ed. Bingley, United Kingdom: Emerald Publishing Group. https://books.emeraldinsight.com/resources/pdfs/chapters/9781848552500- TYPE23-NR2.pdf Farmer, Charles M. 2017. “Relationship of Traffic Fatality Rates to Maximum State Speed Limits,” Traffic Injury Prevention 18 (4): 375–80. https://doi.org/10.1080/15389588.2016.1213821. FHWA (Federal Highway Administration). Crash Modification Factors (CMF) Clearinghouse (database), FHWA, Washington, DC (accessed December 20, 2020), http://www.cmfclearinghouse.org/index.cfm. Graham, Andrew, and P. Sparkes. 2010. “Casualty Reductions in NSW Associated with the 40 km/h School Zone Initiative.” Paper presented at the 2010 Australasian Road Safety Research, Policing and Education Conference, Canberra, Australia, August 31–September 3, 2010. https://acrs.org.au/files/arsrpe/R2010220.pdf. Hauer, Ezra. 1997. Observational Before-After Studies in Road Safety: Estimating the Effect of Highway and Traffic Engineering Measures on Road Safety.” Oxford, United Kingdom: Pergamon, Elsevier Science, Ltd. ITF (International Transport Forum). 2018. Speed and Crash Risk. Paris: Organisation for Economic Co-operation and Development (OECD) Publishing. https://www.itf-oecd.org/sites/default/files/docs/speed-crash-risk.pdf. Jurewicz, Chris. 2010. Road Safety Engineering Toolkit. Sydney, Australia: Association of Australian and New Zealand Road Transport and Traffic Authorities (Austroads). http://casr.adelaide.edu.au/rsr/RSR2007/JurewiczC.pdf or https://acrs.org. au/files/arsrpe/RS07033.pdf. McTiernan, David, Peter Ellis, Robert Grove, and Ganesh Vengadasalem. 2015. Guide to Road Design—Part 1: Introduction to Road Design (AGRD01-15). Sydney, Australia: Austroads. https://austroads.com.au/publications/road-design/agrd10. NSW Centre for Road Safety. (n.d), ‘Safe System’ – the key to managing road safety. Fact Sheet 6. Roads and Traffic Authority of New South Wales. Sliogeris, John, and Roads Corporation. 1992. 110 Kilometre per Hour Speed Limit—Evaluation of Road Safety Effects. Carlton, Victoria, Australia: Victoria Road Safety and Traffic Authority. Turner, Blair, Eric Howard, and Jeanne Breen. 2015. Road Safety Manual: A Manual for Practitioners and Decision Makers on Implementing Safe System Infrastructure. Paris: Permanent International Association of Road Congresses (PIARC). https://roadsafety.piarc.org/en. Webster, D. C., and A. Mackie. 1996. Review of Traffic Calming Schemes in 20mph Zones. TRL Report No. 215, Transportation Research Laboratory (TRL), Wokingham, Berkshire, United Kingdom. Webster, D. C. and R. E. Layfield. 2003. “Review of 20mph Zones in London Boroughs,” Transportation Research Laboratory (TRL) Published Project Report No. PPR243, IHS Markit, Bracknell, Berkshire, United Kingdom. Woolley, Jeremy, Chris Stokes, Blair Turner, and Chris Jurewicz. 2018. Towards Safe System Infrastructure: A Compendium of Current Knowledge. Association of Australian and New Zealand Road Transport and Traffic Authorities (Austroads). https://www.researchgate.net/profile/Chris-Jurewicz/publication/325923186_Towards_Safe_System_Infrastructure_A_ Compendium_of_Current_Knowledge/links/5b2c9fce0f7e9b0df5ba69cb/Towards-Safe-System-Infrastructure-A- Compendium-of-Current-Knowledge.pdf. 18 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? 4. Methodology and Data Sources 19 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS To evaluate the effectiveness of the treatment— Where, “speed limit reductions”—the study team used an observational before-after study with a control group. π = What the expected number of target crashes Following the method by Hauer (1997) , as described would have been in an “after” period without a below, the team obtained counts of crashes before treatment and and after in both treatment site and comparison group, comparing the totals to quantify the benefit λ = The expected number of target crashes in the of the treatment, if any. The team conducted several “after” period at the treated sites different analyses to check the impact of speed limit reductions on the following: 1) the total number of μ = The expected number of target crashes in the crashes, 2) vehicle-to-vehicle crashes, and 3) vehi- “before” period at comparison sites cle-to-pedestrian crashes. ν = The expected number of target crashes in the In an observation before-after study involving a treat- “after” period at comparison sites ment and a comparison group, let K and L denote the corresponding before and after crash counts in The aforementioned hope the following equation can the treatment groups, and M and N are the before be expressed: and after crash counts in the comparison group. The expected values of these crash counts are denoted rΤ=rC or equivalent, that rC=rΤ=1 by the corresponding Greek letters κ,λ,μ, and ν (2) respectively. From the definition of rT it follows that The Comparison Group (C-G) method is based on the hope that in the absence of treatment, the ratio of π=rΤκ the expected number of target crashes “before” and (3) “after” would be the same in the treatment and com- parisons groups. To bring the hope into the orbit of Therefore, if the assumption in Equation (2) is true, explicit analysis, Hauer (1997) distinguished between then it is also true that the ratio rC defined for the comparison group and a parallel but distinct ratio for the treatment group. π=rCκ (4) Thus, defined as: Since rC in equation 4 can be estimated from the rC ≐ ν ⁄ μ to be the ratio of the expected crash counts number of crashes in the comparison group (M and for the comparison group, and N), and κ can be estimated by the number of crashes in the treatment group in the “before” period (K), π rΤ ≐ π ⁄ κ to be the corresponding ratio for the can be estimated. While equation 2 is an assumption, treatment group the only defensible argument put forward by the (1) author was to justify the use of a comparison group 20 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? in an observational study is empirical or inductive. In the simple before-after method π̂=K, whereas in Also, if one can show that in a time series of past the C-G method, π̂=rΤK ̂ . The purpose of rΤ̂ (or of its values rΤ and rC were sufficiently similar then, cog- replacement rC)̂ is to count for the effect of change in nizant of the usual limitations of all inductive argu- various uncontrolled causal factors. ments, one might hope the past similarity also held for that specific value of rC used in a specific C-G To assess the impact of speed limit reductions on study. However, if this is the argument on which the safety performance, the study team used crash data C-G method rests, one must allow in the analysis for from Daegu Metropolitan City (hereinafter Daegu) the possibility that the assumption rC ⁄ rΤ =1 is not from between the years 2013 and 2018. The Korea exactly true in any specific C-G study. It is therefore Transport Institute (KOTI) selected the entire target necessary to consider that ratio rC ⁄ rΤ to be a ran- analysis period, looking at the three years before dom variable which on different occasions takes on (2013~2015) and the three years after (2016~2018), different values. Consistent with the usual statistical during which time Daegu actively pursued speed limit terminology associated the ratio rC ⁄ rΤ will be called changes at all treatment sites. the “odds ratio” and given the symbol ω. For the comparison groups, that is the sites where ω≐rC ⁄rΤ speed limits remained unchanged, a similar duration (5) of three years before (2013~2015) and three years after (2016~2018) was chosen. While the road geom- However, for the purpose of C-G study, Hauer (1997) etry and traffic volumes in treatment and compar- suggested the following steps: ison groups are not identical, they are comparable as reported by KOTI. However, it is important to Step 1: Estimates of parameters: keep in mind treated sections generally experience higher rates of crashes, fatalities, and injuries than ̂ rĈ (N/M)/(1+1/M) ≃ N/M, π̂ =rTK λ̂ = L, rΤ= ̂ untreated sections. In addition, bias is likely in before- the-change crashes in the treated sections, as ran- Step 2: domly allocating sites to the treatment group or the control group is often difficult. ̂ ) ⁄ rT2 = 1/M + 1/N + VÂR{ω}, VÂR{λ} = L,VÂR{rT} ̂ ≃π̂2 [1/K+VÂ R{rΤ} VÂR{π } ̂ /rT2] As mentioned earlier, the Korean National Police Agency adopted the speed management initiative Step 3: known as “Safety Speed 5030” in 2016, and in sev- eral urban areas, the arterial speed limits are set at δ̂ = ̂ =VAR{π̂}+VAR{λ} π-λ, VAR{δ} ̂ 50 kilometers per hour (kph), with side-road speeds set at 30 kph (or sometimes 20 kph) for better Step 4: road safety, as the photos show in figures 4.1 and 4.2. However, as reported by KOTI, not much has θ*=(λ⁄π)/[1+VAR{π̂}/π2], VAR{θ} ̂ ⁄ λ2)+ ̂ ⋍ θ2 [(VAR{λ} changed in road infrastructure; to date, only speed (VAR{π̂}/π )] ⁄ [1+VAR{π̂} ⁄ π ] 2 2 2 21 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS limit signs and road markings have been updated. As limit reductions as candidate regions in terms of the a result, safety assessment in this study primarily current state of speed limit changes. Seoul, Daegu, captures the effects of speed limit change, with- Daejeon, Chungbuk, Jeonnam, Kyeongbuk, Jeju, and out associated traffic calming measures. seven other local police agencies announced speed limit changes on their respective homepages. Table For the selection of target roads, the KOTI team 4.1 shows the results from a survey of the speed limit investigated major cities actively implementing speed reduction sections. Figure 4.1. 30 kph Limit in Residential and Commercial Neighborhoods (with Pavement Marking) a. Residential neighborhood b. Commercial neighborhood Source: Original images captured for this publication. Figure 4.2. Pedestrian Priority Zone with 20 kph Limit and Modified Pavement Texture Source: Original images captured for this publication. 22 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? As shown in table 4.1, Seoul implemented reduced limit reduction sections and diverse speed limit cases. speed limits at a total of 2,534 locations, mostly in Daegu has reduced the speed limits in 865 location protected areas such as school zones; Daegu imple- across 54 roads, excluding those of protection zones, mented limits in 865 locations; Deajeon implemented from 40~80 kph to 30~70 kph. Table 4.2 shows all in 168 locations, and so on. such modified cases with the original speed limit, the lowered speed, limit, and the numbers of roads Of the candidate regions, Daegu was chosen as the where speed reductions have been implemented. target region because of its high number of speed Table 4.1. Current Status of Lowering Speed Limits across Korea Local government Number of Official data Other reduction sections release date Seoul City 2,534 08/31/2016 Speed limit reductions tend to be 30 kph reduction sec- tions due to designation as protection zones Daegu City 865 09/2017 — Daejeon City 168 08/22/2016 — Chungcheongbuk-do Province 253 08/30/2011 — Jeollanam-do Province 754 09/19/2016 — Gyeongsangbuk-do Province 313 08/31/2016 — Jeju-do Province 162 04/2019 — Source: Original calculations produced for this publication. 23 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS Table 4.2. The Current Status of Lowering Speed Limits in Daegu Metropolitan City Case Current speed limit (kph) Modified speed limit (kph) Number of roads Number of control roads 1 80 70 2 1 2 70 60 19 3 3 60 50 18 4 60 40 3 30 5 60 30 3 6 50 40 1 29 7 50 30 5 8 40 30 3 11 Source: Original calculations produced for this publication. Finally, to compare the impact of a speed limit reduc- The three Ulsan road sections include the following: tion—as well as signal delay—on travel time, the (1) the 2.3 kilometer extension section on Hwahop-ro study team selected three different road sections (with a speed limit of 50 kph); (2) the 2.3 kilometer in Ulsan Metropolitan City and measured individual extension section on Samsan-ro (with a speed limit of vehicle speeds to assess the effects of a speed limit 60 kph); and (3) the 2.3 kilometer extension section change on overall travel time. While the ideal data for on Moonsu-ro (also with a speed limit of 60 kph). To such comparison would have been travel time col- enhance the accuracy of the assessment, peak and lected at the before-and-after period from the same off-peak times were separated: 08:00~09:00 a.m. was road sections, such data were not available, hence selected as peak time, and 23:00~midnight as off- influences on speed could only be estimated from peak time. three similar road sections with different speed limits. References Hauer, Ezra. 1997. Observational Before-After Studies in Road Safety: Estimating the Effect of Highway and Traffic Engineering Measures on Road Safety.” Oxford, United Kingdom: Pergamon, Elsevier Science, Ltd. 24 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? 5. Results 25 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS Assessment of Safety Benefit from Speed Limit Reductions This section presents the results of safety assessment Before vs. after comparison of total crashes from reduced speed limits (mostly sign-only as very few infrastructure-based speed management mea- To begin, the study team compared total crash sures have been implemented) in terms of reductions counts for Daegu for the three years before the in crashes, fatalities, and injuries, if any. A reduction change (that is, 2013~2015) with those of the three in vehicle and pedestrian crashes and injuries would years after the change (2016~2018) for the roads be expected as a result of a reduction in average where speed limits have been reduced. These sec- speed and top percentile speeds due to speed limit tions are referred to here as “treatment groups.” The reductions as well as an increase in driver awareness “comparison groups” are roads of similar types where and alertness about speed limit reductions. However, speed limits remained unchanged. For the compari- if speed limit reductions do not result in actual reduc- son groups, the study team chose the same duration tions in the average speed of operation, it might not of before (2013~2015) and the same three years after be effective to simply reduce speed limits, without (2016~2018). associated infrastructure and enforcement to sup- port the new limits. Table 5.1. Before vs. After Comparison of Total Crashes Number of Number of Number of Equivalent property Fatalities crashes fatalities injuries damage only crashes per 100 crashes (EPDO) Before change in speed limit 8,891 114 3,142 19,869 1.28 After change in speed limit 8,794 92 2,852 18,906 1.05 Percent reduction 1.1% 19.3% 9.2% 4.8% 18.4% Sections with unchanged speed limit, –3.0% 6.8% 11.8% 3.0% 9.6% percent reduction Source: Original calculations produced for this publication. Note: A negative (–) sign before the percent change indicates an increase; a positive (+) sign indicates a reduction. A comparison of total crash counts, fatalities, and groups, compared to reductions of 3 percent, 6.8 injuries, as conducted by the Korea Transport percent, and 11.8 percent respectively, in the com- Institute (KOTI) is shown in table 5.1. This indi- parison groups. Note that a negative (–) sign before cates reduced speed limits resulted in a consistent the percent change indicates an increase; a positive decrease in crash statistics across the board: a 1.1 (+) sign indicates a reduction. When compared with percent decrease in total number of crashes, a 19.3 equivalent property damage only (EPDO) crashes and percent decrease in number of fatalities, and a 9.2 fatalities per 100 crashes, in both cases the reduc- percent decrease in number of injuries in treatment tions were higher at treatment sites compared to the 26 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? control sites (similar road types in Daegu). The KOTI estimated following the before-after methodology as also reported results from chi-square tests to check if stated in chapter 4 (Methodology and Data Sources), the reductions were statistically significant; however, with results shown in table 5.2. The shaded columns the results are not promising due to a small sample, with light blue colors correspond to results from the with the total number of crashes showing reductions analysis of total crashes irrespective of road users only at a 14 percent level of significance. and types. The results indicate a 4.2 percent reduc- tion in total crashes, with a standard deviation of 4 However, when checked for effective reductions in percent, and a 15.2 percent reduction in fatalities with crashes using the before-after method with compar- a standard deviation of 28 percent in the treatment ison groups evidence point to reductions for both groups, that is, at the sites with reduced speed limits. total crashes as well as fatalities. The reductions are Table 5.2. Calculation Table for Comparison of Crashes Using the Before-After Method with Control Groups Crash Fatalities Serious Injuries Parameters All crashes Vehicle- Pedestrian All crashes Vehicle- Pedestrian All crashes Vehicle- Pedestrian to-vehicle to-vehicle to-vehicle STEP 1 Find ƛ and π̂ K 8,891 7,194 1,407 114 42 62 3,142 2,321 677 L 8,794 7,049 1,477 92 41 38 2,852 2,045 693 M 3,467 2,568 820 44 11 27 1,305 876 395 N 3,575 2,720 801 38 12 20 1,159 772 369 λ̂=L 8,794 7,049 1,477 92 41 38 2,852 2,045 693 Biased r̂C 1.03 1.06 0.98 0.86 1.09 0.74 0.89 0.88 0.93 Unbiased 1.03 1.06 0.98 0.84 1.00 0.71 0.89 0.88 0.93 rΤ̂= rĈ (N/M)/(1+1/M) ≃N/M π̂ = rΤ̂ K 9,165 7,617 1,373 96 42 44 2,788 2,043 631 STEP 2 Find VÂ R{λ̂} and VAR{π̂} VÂR{λ̂} = L 8,794 7,049 1,477 92 41 38 2,852 2,045 693 ω ́=rC ⁄ rT 1.000 1.000 1.001 1.023 1.091 1.037 1.001 1.001 1.003 VÂR{ω} 0.0008 0.0010 0.0039 0.0687 0.2224 0.1295 0.0023 0.0034 0.0082 Approximation ≃ 1/K + 1/L + 1/M + 1/N VÂR{rΤ̂}⁄rΤ2 = 1/M + 1/N + 0.0014 0.0018 0.0063 0.1177 0.3967 0.2165 0.0039 0.0058 0.0134 VÂR{ω} VÂR{π̂}≃π̂2 [1/K+VÂR{rΤ̂}⁄rΤ2] 123,902 112,203 13,255 1,172 742 456 33,007 25,983 5,922 27 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS Crash Fatalities Serious Injuries Parameters All crashes Vehicle- Pedestrian All crashes Vehicle- Pedestrian All crashes Vehicle- Pedestrian to-vehicle to-vehicle to-vehicle STEP 3 Find δ̂ and θ̂ VÂ R{π̂}/π2 0.001 0.002 0.007 0.126 0.420 0.233 0.004 0.006 0.015 δ̂=π-λ 371 568 –104 4 1 6 –64 –2 –62 θ*=(λ⁄π)/[1+VAR{π̂}/π2] 0.958 0.924 1.068 0.848 0.687 0.696 1.019 0.995 1.082 STEP 4 Find the estimate of VAR{δ̂} and VAR{θ̂} σ̂{δ̂} 364 345 121 36 28 22 189 167 81 σ̂{θ̂} 0.04 0.04 0.09 0.28 0.32 0.29 0.07 0.08 0.14 Results Reduction 371 568 –104 4 1 6 –64 –2 –62 Standard deviation 364 345 121 36 28 22 189 167 81 Amounts to a x% reduction 4.2% 7.6% –6.8% 15.2% 31.3% 30.4% –1.9% 0.5% –8.2% With a standard deviation of 4% 4% 9% 28% 32% 29% 7% 8% 14% Source: Original calculations produced for this publication. Before vs. after comparison of vehicle-to-vehicle crashes Following the same method, though using only vehi- number of injuries by 11.9 percent when comparing cle-to-vehicle crashes for Daegu, the study compares absolute numbers. However, when compared with crash records from three years before (2013~2015) the control sites, the study team observed both the and three years after (2016~2018), along with data for number of crashes and the number of fatalities had the same duration for the comparison groups. Results increased at the control sites, but with no decrease in in table 5.3 show a decrease in crashes by 2.0 percent, the number of crashes resulting in injuries. in the number of fatalities by 2.4 percent, and in the 28 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? Table 5.3. Before vs. After Comparison of Vehicle-to-Vehicle Crashes Total crashes Number of Number of Number of Equivalent property Fatalities (Vehicle-to vehicle) crashes fatalities injuries damage only crashes per 100 crashes (EPDO) Before change in speed limit 7,194 42 2,321 15,737 0.58 After change in speed limit 7,049 41 2,045 15,002 0.58 Percent reduction 2.0% 2.4% 11.9% 4.7% 0.4% Sections with unchanged speed limit, –5.9% –9.1% 11.9% 0.1% –3.0% percent reduction Source: Original calculations produced for this publication. Note: A negative (–) sign before the percent change indicates an increase; a positive (+) sign indicates a reduction. When checking results from table 5.2 for estimated period” (2016~2018) for the treatment groups. Sim- actual reductions in the vehicle to vehicle crashes, ilar to other crash types, the analysis period for the fatalities, and injuries—keeping the crash trend at the unchanged speed limit sections, that is, for the “con- control groups in mind—clear evidence emerges of a trol groups” consisting of similar road types, were reduction in the total number of the vehicle to vehi- also the same, with three years before (2013~2015) cle crashes. KOTI estimates using chi-square values and the same three years after (2016~2018). A com- also indicate statistically significant reductions of total parison of the total crash counts, fatalities, and vehicle-vehicle crashes at a 5 percent level of signifi- injuries shows the reduction in speed limit resulted cance. The results from before-after using the com- in a significant reduction in fatalities, though slightly parison group for these crash types are presented increased the total number of crashes and injuries in columns with light orange colors and the esti- under this crash type. mates show a 7.6 percent reduction in vehicle-vehicle crashes with a standard deviation of 4 percent. There As reported by KOTI and shown in table 5.4, the is also evidence of reductions in fatalities, with a 31.3 speed limit reductions resulted in an increased percent reduction in numbers while controlling for pedestrian crashes by 5.0 percent, and injury crashes reductions in the comparison group and with an esti- by 2.4 percent, while fatalities decreased by 38.7 per- mated 32 percent standard deviation of the reduction. cent when an absolute number of vehicle-pedestrian crashes are considered in treatment groups, without Before vs. after comparison of considering the information from the comparison vehicle-to-pedestrian crashes groups. Additionally, the EPDO coefficient reduced by 4.8 percent. Notably, in the control group, results Vehicle-to-pedestrian crashes for Daegu from the show a decrease in crashes of all the categories, but three-year “before period” (2013~2015) are com- at 5 percent level of significance, none of the differ- pared with crashes during the three-year “after ences is statistically significant. 29 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS Table 5.4. Before vs. After Comparison of Vehicle-to-Pedestrian Crashes Total crashes Number of Number of Number of Equivalent property Fatalities (Vehicle-pedestrian) crashes fatalities injuries damage only crashes per 100 crashes (EPDO) Before change in speed limit 1,407 62 677 3,447 4.41 After change in speed limit 1,477 38 693 3,280 2.57 Percent reduction –5.0% 38.7% –2.4% 4.8% 41.6% Sections with unchanged speed limit, 2.3% 25.9% 6.6% 8.8% 24.2% percent reduction Source: Original calculations produced for this publication. Note: A negative (–) sign before the percent change indicates an increase; a positive (+) sign indicates a reduction. Finally, when checked for effective reductions While the results from the overall safety assessment in pedestrian crashes due to speed limit reduc- indicate some benefits from speed limit reductions, tions using the before-after method with compar- in the absence of detailed data on operating speed ison groups, evidence shows pedestrian fatalities before and after, the compliance rate of such speed decreased in the treatment groups compared to the limit reductions cannot be confirmed. Further, reduc- decrease in pedestrian crashes in the control groups. tions in crashes do not follow expectations, even The results indicate the reduced speed limits saved when the overall decreasing trend of crashes, fatali- the lives of six pedestrians, which translates to a 30.4 ties, and injuries in Daegu are taken into account with percent reduction with a standard deviation of 29 the help of “comparison groups.” percent as shared previously in table 5.2 (light green color). However, results show no reductions in total pedestrian crashes and injuries. 30 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? Assessment on Impact of Travel Time from Speed Limit Reductions Table 5.5 shares results from the analysis of travel time due to two different speed limits as reported by KOTI. For the Hwahop-ro section with 50 kph speed limits, the nighttime average travel time was 285.0 seconds, while the morning peak travel time was 369.4 seconds. For Samsan-ro, nighttime average travel time was 249.3 seconds, while morning peak travel time was 313.5 seconds. For Moonsu-ro, nighttime average travel time was reported to be 289.8 seconds, while morning peak travel time was 643.0 seconds. Table 5.5. A Comparison of Travel Time across Study Sections in Daegu Metropolitan City Type Hwahop-ro Samsan-ro Moonsu-ro Speed limit Speed 50 kph Speed 60 kph Speed 60 kph Section extension 2.3km 2.3km 2.3km Number of lanes 2 lanes 4 lanes 3 lanes Free flow time (seconds) 166 sec 138 sec 138 sec Nighttime 285.0 sec 249.3 sec 289.8 sec Travel time Morning peak 369.4 sec 313.5 sec 634.0 sec Intersection Nighttime 119.0 sec (41.7%) 111.3 sec (44.6%) 151.8 sec (52.4%) delay time Morning peak 203.4 sec (55.1%) 175.5 sec (56.0%) 496.0 sec (78.2%) Average number of inter- Nighttime 1.0 times 0.6 times 1.9 times section delays per vehicle Morning peak 1.7 times 1.3 times 3.8 times Average travel speed for Nighttime Speed 46.0 kph Speed 60.1 kph Speed 56.7 kph one-way road Morning peak Speed 43.3 kph Speed 49.8 kph Speed 34.1 kph Source: Original calculations produced for this publication. Of the total travel times, the Hwahop-ro section shows the lowest proportion of delay time for both the night- time and the morning peak. In turn, this indicates the efficiency of mobility is highest in the Hwahop-ro sec- tion, despite the lower speed limit of that road compared to the two other test sections. While these promising results clearly indicate a higher speed limit does not necessarily translate into quicker travel time, collecting travel times before and after the change of posted speed limits from a particular corri- dor would also have been beneficial, as travel times are key performance indicators, and stand as a major hur- dle to overcome while advocating for reduced speed limits. However, the study results clearly show a higher speed limit could very well result in greater delays and longer travel times. 31 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS 6. Conclusions and Policy Recommendations 32 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY? Speed is undeniably a key risk factor for road and injuries. However, road development in LMICs crash-related injuries and fatalities. Therefore, ensur- can benefit from lessons learned in other countries, ing operational speeds are safe for all road users especially the lesson of safer mobility by managing demands careful speed management driven by speed on newly developed and/or upgraded roads. accurate data and supported by a process that yields Korea’s experience from speed limit reductions, success in other transportation environments. A first which has likely resulted in reduced average speeds, step in effective speed management is to discourage along with additional benefits to road safety, could be speeding through the use of appropriate speed limits. very similar to those observed in other studies in the United Kingdom, Europe, and Australia. However, an However, garnering support from policy makers—and evaluation conducted in Asia and collaboration with even transport professionals—for reducing speed the World Bank could be more acceptable to neigh- limits is often an uphill battle for several reasons. boring Asian countries and developing countries in First, neither group is fully convinced about the cause- other regions. and-effect of speeding and injuries and fatalities, and thus the immediate safety and health cost benefits of Case in point: When the Korean National Police reducing speed. Second, they demonstrate an unwill- Agency initiated “Safety Speed 5030”—setting urban ingness to believe the research data indicating lower arterial speed limits at 50 kilometers per hour (kph) speed limits can actually improve traffic mobility, and side-road speed limits at 30 kph (and in some rather than creating longer travel times. areas 20 kph)—ordinary Korean citizens did not organize major protests or voice mass disapproval Third, many share a common misperception that the against the initiative. However, the national police public pushes to drive at ever higher speeds—which agency did launch a widespread campaign to win is a myth. In reality, an authoritative survey, the E-Sur- the hearts and minds of the Korean public and help vey of Road Users’ Attitudes (ESRA), conducted by the sensitize them to the multiple and wide-ranging ben- Brussels-based Vias institute, indicates fewer than 20 efits of the speed limit reduction initiative. “You can percent of road users find it acceptable to drive faster see people, once you reduce speed” was the slogan than the speed limit; fewer than 10 percent indicate widely promoted by means of video clips, posters, that it is acceptable to speed in built-up areas; and and other public relations materials, and the results most (up to 80 percent) believe speed is a cause of of an experiment in Busan was circulated as a press road crashes. Additionally, 90 percent of respondents release. The experiment demonstrated the total suggest traffic rules should be stricter.1 savings in travel time was only 2 minutes when two travelers maintained 50 kph and 60 kph while driving As motorization, road development, and travel the identical routes of 15 kilometers long. Once the speeds continue to increase in low- to middle-income Korean public understood the issue, the speed limit countries (LMICs), so do the road crashes, fatalities, reduction initiative received support. 1 To learn more about the E-Survey of Road Users’ Attitudes (ESRA) coordinated by the Vias Institute in Brussels, Belgium, visit the ESRA website: https://www.esranet.eu. 33 LESSONS FROM THE REPUBLIC OF KOREA’S RECENT MOVE TO LOWER SPEED LIMIT ON URBAN ROADS Findings from the case study conducted in Daegu benefits. However, such engineering treatments to show a significant reduction in total and vehi- implement active measures of speed control have cle-to-vehicle crashes in the treated sites, and an been limited in this current study. While the impact of overall reduction in fatalities, with six lives saved from speed limits combined with supporting engineering vehicle-pedestrian crashes due to the treatment over treatments on road safety is significant, the study did the three-year period. However, this study is also not not have enough sample to analyze these effects in without limitations: First, no robust data collection Korea. It is, therefore, strongly recommended to pri- was done before and after speed limit reductions to oritize installing traffic calming measures on the road check speed compliance and the actual reductions in with a speed limit less than or equal to 30 kph. Even average speeds. As a result, the effect of speed limit with these limitations, this KOTI-World Bank joint reductions on average operating speed is largely study begins to fill that gap with far-reaching implica- unknown for this study. Further, research evidence tions with the potential to initiate policy dialogue on shows different speed limit reductions will result in speed limit reductions in developing countries in Asia different reductions in average speeds. As a result, it and elsewhere. would have been more scientific to conduct separate safety assessments for roads with different speed Lessons for World Bank operations: With increased limit reductions. However, this was not possible due emphasis on road safety in World Bank operations, to the small sample sizes with lower-level confidence, speed management and providing engineering which needs to be studied further to determine measures to support context-sensitive operating the long-term impact of the speed limit reductions. speed has become increasingly important. Task team Regardless of the study’s shortcomings, the results leaders (TTLs) also need to appreciate the safety from the travel time analysis prove particularly prom- benefits of appropriate speed limits and other associ- ising, indicating a higher speed limit could very well ated measures to increase speed compliance result- result in greater delays and longer travel times. The ing in better safety outcomes for their projects. It is practical implication of this is that concerns about thus very important for TTLs to adopt good design longer travel times are misplaced and should not practices, along with measures of performance, to prevent transportation planners and regulators from achieve the desired level of safety outcomes with lowering speed limits to enhance road safety. intermediate outcome indicators for their projects. For this purpose, gathering all relevant base data is Based on the earlier studies, and from this research, essential as part of the project preparation, to sup- the study team concludes that in general, speed port a rich and useful result framework. From that limit reductions—supported by strong visuals and point of view, the experience of KOTI-WB collabora- engineering treatments—go further in generating tion will guide Word Bank TTLs to help achieve the substantial compliance and subsequently, the safety desired project results. 34 DO SPEED LIMIT REDUCTIONS HELP ROAD SAFETY?