Transport Policy 29 (2013) 192–198 Contents lists available at SciVerse ScienceDirect Transport Policy journal homepage: www.elsevier.com/locate/tranpol Some measures for sustaining red-light camera programs and their negative impacts Qiang Yang, Lee D. Han n, Christopher R. Cherry Department of Civil & Environmental Engineering, The University of Tennessee, 112 Perkins Hall, Knoxville, TN 37996, United States art ic l e i nf o Keywords: Red-light camera Red-light running Revenue Signalized intersection Traffic safety Law enforcement a b s t r a c t Automated enforcement red-light cameras (RLC) have been widely adopted by municipalities around the world as a measure of curbing red-light running (RLR) at signalized intersections and reducing the cost of law enforcement. While a consensus has not yet been reached about whether RLC in general can benefit intersection safety by reducing RLR and crashes, recent debates revolve around using RLC as a revenue generator. Some of the political backlash of RLC is the perception that they are installed primarily to fulfill revenue guarantees and sustain the RLC program. Some municipalities have been charged with changing the signal phasing to trap more red-light runners and increase the revenue from RLC programs. This paper focuses on a number of engineering strategies, mainly related to signal timing that may be used by municipalities to achieve their financial goals. The negative impacts of implementing these measures on the safety and efficiency of intersection operations and public support on RLC programs are also discussed. These strategies are also revealed to increase transparency of the divergent motivations of RLC vendors, municipalities, policy makers and safety advocates. Published by Elsevier Ltd. 1. Introduction Since invented in 1960s, automated enforcement red-light cameras (RLC) have been widely adopted by municipalities around the world as a measure to curb red-light running (RLR) at signalized intersections and reduce the cost of law enforcement. Technological improvements have made RLC much more effective in recent years, increasing their adoption in the past decade. A 2009 study revealed that about 350 communities in the US used RLC (Chatterjee and Cate, 2009). While a consensus has not yet been reached about whether RLC can benefit intersection safety by reducing RLR and crashes, municipalities are facing a major ethical dilemma around balancing financial guarantees to sustain RLC programs and improving safety. A literature survey reveals that most municipalities implementing RLC are committed to private RLC providers with certain revenue goals to financially sustain their RLC programs. Most RLCs are installed with dual, conflicting purposes, reduce RLR and maximize private (and public) sector revenue from RLR citations. Harmonizing these two purposes is challenging resulting in substantial backlash against RLC. Indeed, as municipal budgets are threatened, the temptation to identify RLC as a revenue generation source is increasing. To achieve their revenue goals, some municipalities have implemented certain n Corresponding author. Tel.: +1 865 974 7707; fax: +1 865 974 2669. E-mail addresses: qyang5@utk.edu (Q. Yang), lhan@utk.edu (L.D. Han), cherry@utk.edu (C.R. Cherry). 0967-070X/$ - see front matter Published by Elsevier Ltd. http://dx.doi.org/10.1016/j.tranpol.2013.06.006 engineering measures to trap red-light runners at RLC equipped intersections and increase the citations/revenue from RLC programs. This paper begins by giving a background on the financial and policy issues related to RLC programs. Next, the key focus of this paper is to present various engineering measures that may be employed by municipalities to increase the RLR and revenue. The measures discussed in this paper are mostly related to the signal timing that is relatively inexpensive to implement. The implementation of these measures affects the safety and efficiency at intersections. Meanwhile, protests and lawsuits concerning the use of RLC have occurred around the United States and, in many instances, lawmakers have restricted their use. The paper discusses the ethical challenges and negative impacts of these measures on intersection safety and efficiency and the credibility and public image of related agencies and elected officials. 2. Background 2.1. Effect of RLC on intersection safety The primary motivation for installing RLC is safety improvement through the consistent expectation of enforcement. As such, a number of studies have evaluated the effectiveness of RLC as an enforcement mechanism to reduce red-light violations and associated severe crashes. Several studies found a significant difference in crash rate and an improvement in overall safety attributable to RLC. Retting et al. (1999a) and Ruby and Hobeika (2003) Q. Yang et al. / Transport Policy 29 (2013) 192–198 investigated the RLC in Virginia and found a 36% and a 69% reduction in RLR over the first three and six months of camera operation. The crash rates were also reduced by 40%. Similar RLC positive effects on RLR and crash rates were observed in California (Fleck and Smith, 1999; Retting et al., 1999b), North Carolina (Cunningham, 2004), and Iowa (Fitzsimmons et al., 2009). Lum and Wong (2003a,b,c) found that the RLC installation at three intersections in Singapore reduced RLR by more than 40% while non-camera approaches did not experience such a reduction during the same period. Huang et al. (2006) modeled the crash risk at 15 signalized intersections in Singapore and found that RLCs were effective in reducing RLR and right-angle collisions. However, it had a mixed effect on rear-end collisions depending on the speed of the trailing vehicle and the headway between vehicles. Persaud et al. (2005) reported a similar effect of RLC on right-angle and rear-end collisions in the US. Radalj (2001) investigated the same issue at 58 RLC and 447 non-RLC intersections in Australia and found that the installation of RLCs reduced fatalities by over 50% but increased rear-end crashes by 17%. The reduction in the total number of crashes was 3%. Some other studies found no significant difference or even negative effects on safety after the installation of RLC. Burkey and Obeng (2004) analyzed crashes occurring near 303 intersections over a 57-month period. They found RLCs increased the crash rates by 40% while the overall trend during the same period indicated that crashes at all intersections were becoming less frequent. The study reported a large increase in rear-end crashes due to RLCs. Regarding crash severity, RLCs were found to increase property damage only and possibly injury crashes, but have insignificant effect on severe crashes. A study in Arizona (Washington and Shin, 2005) found that the total number of crashes were unchanged as a result of RLCs at 10 intersections in the City of Phoenix (14% reduction in angle crashes and 20% increase in rear-end crashes). Total crashes were reduced by 11% in the City of Scottsdale. Garber et al. (2007) also observed an increase in rear-end crashes and a reduction in RLR crashes associated with RLCs. However, when the comprehensive costs for different types of crashes were monetized, RLCs were associated with a net increase in crash costs considering six jurisdictions in Virginia. Kent et al. (1995) investigated RLR data at three RLC intersections and concluded that there was no difference in RLR between camera and non-camera approaches (at the same intersection). 2.2. Financial promise of RLC programs Many believe that transportation agencies and vendors install RLCs for the purpose of increasing revenue. Before exploring public responses to RLC programs and engineering measures employed to increase RLC revenue, the background of how RLC programs are funded and sustained is given. Chatterjee and Cate (2009) reviewed various RLC programs in the US and interviewed individuals who were responsible for the implementation and operation of RLC programs in the city of Knoxville and Chattanooga, TN and Baltimore, MD. They reported that the installation and maintenance of RLC devices were provided by private vendors with no cost to cities. Cities were responsible for administration of these programs. They also reported that several North Carolina cities had to discontinue their RLC programs since the state law required 90% of the revenue to be used for local schools, which resulted in the inability of these cities to sustain RLC programs. Garber et al. (2005) investigated the fiscal feasibility of RLC programs in six jurisdictions in Virginia. The report documented the detailed funding mechanism and revenue/cost conditions for these RLC programs. The majority of the RLC programs were provided by private vendors in the form of rental. In return, agencies paid the vendors a flat rate per month or 193 a certain proportion of the citations. Overall, three of the six jurisdictions showed revenue/cost ratio below 1 while the other three were slightly above. Due to the decrease of citations after implementing RLCs, the authors recommended to increase the RLC fine from $50 to $100 to improve the financial sustainability. Maccubbin et al. (2001) conducted a literature review on RLC program contracting mechanisms and the fines and penalties associated with the programs in 17 US cities revealing similar private/public cost and revenue mechanisms. Given that RLC programs are mostly supported by revenues generated from RLC citations and revenue guarantees are usually contracted between private camera providers and municipalities, traffic engineers are facing an ethical dilemma balancing revenue generation to sustain their RLC programs and traffic safety/ efficiency goals. Traffic engineers may have seldom been in such a situation across so many cities. 2.3. Policy response to RLC Political backlash has occurred in many cases because the perception that RLCs are used to generate revenue. Olson (2010) reported the referendum in Houston to cease the use of RLC. A survey of news articles on theNewspaper.com highlighted many anti-camera referendums in cities in Texas, Washington, Missouri, California, and Illinois between 2010 and 2012. States like Massachusetts, South Dakota, Mississippi, Maine, Nevada, Virginia, Alabama, Kentucky, and cities, like Albuquerque, NM and San Jose, CA, even voted to reject the use of RLC. In regions where RLCs were implemented, many lawsuits have been filed challenging the use of RLCs. The Insurance Institute for Highway Safety (IIHS, 2010) reported lawsuits resulted from both citizens and RLC vendors against municipalities. In 2009, a number of cities were sued for the installation of RLC in Florida, where automated enforcement was still illegal in the state at the time of installation. To enable the use of RLCs, these cities created their own ordinance, which was not allowed by the Florida Constitution (Florida Statute Chapter 316) and became the primary argument in the lawsuit against the cities (Ovalle, 2012; Naples News Daily, 2010). Similar cases questioning RLC legality were reported in Minneapolis, MN; Hazelwood, MO; Lafayette, LA; Miami-Dade County, FL; Santa Ana and South San Francisco, CA; and Clive, IA since 2007 (theNewspaper.com, 2009). As a result, some of the illegally collected fines had to be refunded to drivers. Since 2005, many cities shut down their RLC system after a few months or years of operation mainly because of (1) public pressure and legality, (2) failure to generate adequate revenue, and (3) failure to improve safety. Pinkerton (2010) reported the shut-down of the RLC program in Houston, prematurely ending a contract with the private RLC vendor. Atlanta is likely to join Los Angeles, CA and Houston, TX as major cities that have recently discontinued photo ticketing programs (theNewspaper.com, 2012). Many other cities in Georgia, California, Colorado, Washington, Missouri, North Carolina, and Texas had similar experiences. Since some municipalities dropped their RLC program in the middle of contract with camera vendors, which led to the loss of revenue for the vendor, there are also legal disputes between municipalities and RLC vendors. A state law in Tennessee took effect in 2011 that prohibited the use of cameras to issue tickets for right-turn-on-red violations (Tracy et al., 2011). As a result, the RLR citations decreased by three quarters diminishing revenue to both cities and camera vendors. Two RLC vendors filed lawsuits against the city of Knoxville and the town of Farragut respectively after the new state restriction was issued in 2011 (Brewer and Jacobs, 2011). A RLC company also filed a lawsuit against the city of Houston for breach of contract after a 2010 referendum shutting down RLCs. Recently the company has reportedly agreed to drop the lawsuit 194 Q. Yang et al. / Transport Policy 29 (2013) 192–198 against the city and take down all cameras providing the city pays a settlement of at least $4.8 million (Moran, 2012). Similar cases in Manatee, FL; San Bernardino, CA; and Baytown, TX were reported as well (theNewspaper.com, 2009, 2012). To generate revenue from RLC, some local agencies are accused of taking measures to increase the probability of RLR through signal timing or other engineering strategies. Experiences in Chattanooga (Lazenby, 2008) and Nashville, TN (Tobia, 2006); Dallas (KDFW-TV, 2007) and Lubbock, TX (KCBD-TV, 2007); Springfield, MO (Springfield News-Leader, 2007); and Union City, CA (Shatzman, 2005), revealed engineers accused of shortening the yellow light duration at RLC equipped intersections presumably to increase RLR. As a result, the city of Chattanooga in Tennessee (Lazenby, 2008) and the city of Costa Mesa in California (Martinez, 2004) refunded fines to motorists who received tickets for running red-lights at improperly timed intersections. The State of Georgia issued new law mandating longer yellow phases at intersections where RLCs were to be installed (Georgia General Assembly, 2008). The state of Tennessee also proposed legislation to ban cities with RLC from shortening yellow signal duration (Tennessee General Assembly, 2008). Whether or not the municipalities' motives were explicitly trying to increase revenue, it is clear that there is a strong perception that engineering strategies are used to increase RLR and policy is being developed to combat poor engineering practices in the context of RLCs. Table 1 summarizes cases of policy response associated with the use of RLC in different states or municipalities. There are abundant examples of municipalities and vendors accused of adjusting signal cycles to presumably increase revenue. The length of the yellow duration is the primary parameter being adjusted to boost revenue. This paper studies different possible engineering measures, mainly related to the signal timing, which may influence the frequency of RLR and thus could be potentially used by RLC providers and municipalities for increasing revenue from the camera systems. These strategies are also revealed to policy makers and citizens to increase transparency of the divergent motivations of RLC vendors, municipalities, and safety advocates. Table 1 Summary of policy responses associated with the use of RLC. States of municipalities Policy responses Texas, Washington, Missouri, California, and Illinois States: Massachusetts, South Dakota, Mississippi, Maine, Nevada, Virginia, Alabama, and Kentucky. Cities: Albuquerque, NM and San Jose, CA Temple Terrance, FL; Minneapolis, MN; Hazelwood, MO; Lafayette, LA; Miami-Dade County, FL; Santa Ana and South San Francisco, CA; and Clive, IA States: Georgia, California, Colorado, Washington, Missouri, North Carolina, and Texas. Cities: Atlanta, GA;Los Angeles, CA; and Houston, TX Knoxville and Farragut, TN; Houston and Baytown, TX; Manatee, FL; and San Bernardino, CA Chattanooga and Nashville, TN; Dallas and Lubbock, TX; Springfield, MO; Union City, CA Chattanooga, TN and Costa Mesa, CA Anti-camera referendum Georgia, Tennessee Vote to reject the use of RLC Lawsuit against the illegal installation of RLC RLC system shut-down Legal dispute between RLC vendors and municipalities Accused of cheating with the yellow duration for trapping red-light runners Refund ticketing fines due to improper signal timing Law against shortening the yellow duration at RLC intersections 3. Measures This section introduces engineering measures that can potentially increase the frequency of RLR at signalized intersections. The effectiveness of these measures is analyzed either quantitatively or qualitatively. Potential issues related to safety and efficiency of the traffic system and public image of relevant agencies due to implementing such measures are also discussed. 3.1. Shorten yellow duration and/or lengthen all-red duration The length of total change interval (i.e. yellow plus all-red) affects the safety of traffic operation at signalized intersections. But RLR behavior is more associated with the yellow duration part. Engineering measures that tweak the two metrics can affect RLR or safety or both at the same time. 3.1.1. Shorten yellow duration ITE recommends yellow duration to be calculated with Eq. (1) (ITE, 1989). Y ¼ tr þ V 2a þ 2gG ð1Þ where Y is recommended yellow duration, s; tr is driver perception-reaction time, 1.0 s; V is vehicle approaching speed, which is typically the 85th percentile design speed, ft/s; a is deceleration rate, 10 ft/s2 (3.0 m/s2); g is gravitational acceleration, 32.2 ft/s2 (9.8 m/s2), G is approach grade, ft/ft. The yellow duration equation is based on the rationale that drivers at the design speed can cross the stop line within yellow if the distance to the stop line at the onset of yellow is shorter than a comfortable stopping distance. This ensures that drivers at the design speed have one reasonable choice from either comfortably stopping or crossing without running a red-light depending upon their distances to the stop line. In other words, if the yellow duration is shorter than the recommended value, some drivers at a certain distance from the stop line face a dilemma situation and have to run red-light because they are unable to comfortably stop before the stop line. The dilemma zone, which affects the RLR behavior, is discussed in detail in a later section. Bonneson and Zimmerman (2004b) conducted a before-and-after study on the effects of increasing the yellow interval on the frequency of RLR, and found that an increase of the yellow duration by 0.5 to 1.5 s decreased the red-light violations by at least 50% (see Fig. 1). Increasing a yellow interval that is shorter than the ITE recommended value yielded the greatest return. Van der Horst and Wilmink (1986) also reported that a 1.0 s increase in yellow (i.e. from 3.0 s to 4.0 s in urban areas and 4.0 s to 5.0 s in rural areas) decreased red-light violations by 50%. Retting et al. (2008) conducted a similar study of increasing yellow intervals from 3.0 s to 4.1 s and from 4.0 s to 4.9 s at two intersections respectively, and observed a 36% reduction in red-light violations. Bonneson et al. (2002) performed a study in the opposite direction. They studied the effects of decreasing the yellow interval by 1.0 s and reported a 110% increase in RLR violation frequency. At the same time, they observed a 53% reduction in the violation frequency when the yellow interval was increased by 1.0 s. These studies show that shortening the yellow interval duration increases the occurrence of red-light violations and thus revenue generated from RLC, especially when the yellow interval is shortened to a value below the ITE recommended value. One of criticisms of strategies to increase or decrease the yellow time is that drivers may adapt to the change of yellow duration after the measure is implemented. The long-term effect is smaller than the period immediately following the implementation of the measure. Such an effect is called “habituation”. However, studies showed that the habituation effect did not undo Q. Yang et al. / Transport Policy 29 (2013) 192–198 195 However, some studies had different results regarding safety effects of installing or increasing the length of clearance intervals. Bonneson and Zimmerman (2004a) stated that increasing clearance interval was likely to reduce right-angle crashes at the initial onset of red phase (i.e. the first few seconds of red). However, these initial onset-of-red crashes were relatively infrequent so increasing clearance interval might not significantly reduce the total number of rightangle crashes. Souleyrette et al. (2004) evaluated long-term safety effects of increasing clearance intervals employing a before-and-after analysis of 11 years of data and found short-term reductions in crash rates (one year after the implementation), but long-term reductions were not observed. A study in Indiana also reached the same conclusion (Roper et al., 1990). It is likely that drivers may become accustomed to the change of clearance interval after a certain period of time, which could diminish benefits of clearance interval. Fig. 1. Probability of stopping as a function of travel time and yellow duration (Bonneson and Zimmerman, 2004b). the effect of changing the yellow duration (Bonneson and Zimmerman, 2004b; Retting and Greene, 1997; Gårder, 2004). Although ITE recommended yellow duration values are not mandatory (i.e. code or standard), decreasing the yellow duration to an unreasonable low value for the purpose of trapping more red-light runners may impose risks, especially rear-end and rightangle collisions, on drivers. As a result, related traffic departments have to take the responsibility for improper settings of traffic signal timing. Retting et al. (2002) estimated the potential crash effects of modifying the duration of traffic signal change intervals to ITE recommended values. They found an 8% reduction in crashes, a 12% reduction in injury crashes, and a 37% reduction in pedestrian and bicycle crashes at experimental sites relative to controls for a 3-year period following the implementation of signal timing changes. Based on a meta-analysis of all worldwide studies in 1997, Gårder (2004) found that longer evacuation times (all-red and/or yellow times) on average reduced crashes by 55%. Bonneson and Zimmerman (2004a) also reported that crash frequency decreased with increasing yellow duration. 3.1.2. Lengthen all-red duration Reducing yellow duration will increase RLR. A supplemental strategy to counter the safety impacts of reduced yellow time is adding an adequate clearance interval (all-red phase). This strategy effectively compensates yellow time with all-red time. To ensure safety of intersection operation, the total length of change interval (i.e. yellow plus all-red) should be guaranteed. Therefore, lengthening clearance interval will not undo the effects of trapping more red-light runners after shortening yellow interval, and also not cause burden on related agencies for increasing crash rates. Zador et al. (1985) conducted a study at 91 signalized intersections and found that intersections with more adequate clearance intervals had substantially fewer rear-end and right-angle crashes, but yellow running occurrences were unaffected by the length of clearance intervals. Seyfried (2004) also reported that a clearance interval and its length did not influence the drivers' decisions on running red lights. Awadallah (2009) proposed a theoretical approach for reducing RLR and found that clearance intervals were not necessarily a cure for RLR, especially when drivers came to expect an additional safety increment and tried to misuse it. Stein (1986) reported that intersections that had inadequate clearance intervals had higher crash rates. McGee (2003) summarized several studies on the effect of adding a clearance interval on intersection crashes and concluded that all these studies showed positive safety benefits in terms of reducing either crash rates or injuries after implementing clearance intervals. 3.2. Shorten cycle length Shortening the cycle length can increase the red-light violation frequency since shorter cycle lengths increase the hourly frequency of signal changes and thus the exposure of drivers to potential RLR situations. Previous research indicated that the frequency of RLR is largely affected by the frequency with which yellow is presented (Van der Horst and Wilmink, 1986). Many other researchers also recognized the correlation between the frequency of signal changes and RLR (Porter and England, 2000; Baguley, 1988). Based on field studies, Bonneson et al. (2002) found a 29% increase in red-light violations when cycle length was decreased from 90 to 70 s and an 18% reduction in red-light violations when cycle length was increased from 90 to 110 s. In addition to the increase of the frequency of signal transitions, a shorter cycle length affects RLR because it increases the probability of a transition during the end of a vehicle platoon. A very long cycle can fully discharge the vehicle platoon before the next transition and thus reduce the probability of catching violations during the transition. Vehicles arriving in a platoon (or in a close car-following condition) are more likely to run a red-light, which will be further discussed in a later section. Signal timings that promote platoon progression through coordination as well secondary coordination (Sun et al., 2011) can thus be adversely impacted by shortened cycle length. The efficiency of signalized intersection operation is associated with the setting of cycle length. According to Han and Li (2007), relatively short cycle lengths may be desirable if traffic demand is low. However, if traffic arrival rate is higher than the discharging capacity of the short signal cycle, some drivers need to wait in queue for more than one cycle. The excessive delay may also encourage drivers to run a red-light. Therefore, although shortening cycle length increases the occurrence of RLR, it may not always be desirable for operations. 3.3. Increase speed limit Drivers at high speed are more likely to be caught in dilemma zone when approaching a signalized intersection and have a higher probability of running a red-light. When drivers are within a distance from the intersection that is shorter than the distance to comfortably stop and longer than the distance to cross the stop line before the onset of red at a constant speed, they are assumed to be caught in the dilemma zone. Assuming level ground, Eqs. (2) and (3) show the calculation of the necessary stopping distance and the travel distance at a constant speed during yellow. S1 ¼ t r V þ S2 ¼ V  Y V2 2a ð2Þ ð3Þ 196 Q. Yang et al. / Transport Policy 29 (2013) 192–198 Distance to intersection (ft) d 500 400 300 200 100 0 0 10 20 30 40 50 60 Speed (mph) Fig. 2. Intersection dilemma zone at different speeds. etc. Garber and Gadiraju (1989) indicated that drivers tended to operate at increasing speeds as roadway geometric characteristics improved regardless of the posted speed limit. Therefore, intersections with certain roadway geometric conditions (e.g. more lanes, wider lane, lower curvature, etc.) may have a higher RLR frequency. Installing RLC at these intersections may increase revenue generation. In addition, posted speed limit and roadway geometric are interrelated in affecting the operating speed and its variation. Garber and Gadiraju (1989) found that when posted speed limit was approximately consistent with design speed (or roadway geometric conditions), e.g. the posted speed limit is between 6 and 12 mph (9.7 and 19.3 km/h) lower than the design speed, speed variance was minimized. Speed variance increased with increasing difference between the two. Therefore, raising speed limit may also increase the speed variance and the number of vehicles operating in extreme speed. It increases the number of vehicles in the more extreme area of dilemma zone and thus the probability of RLR. However, drivers' indecision in the dilemma zone due to the increase of posted speed limit may lead to aggressive crossing or abrupt stopping behavior. The increased speed limit may also lead to drivers' speeding behavior which is incompatible with actual driving conditions, like road geometry, ambient environments, etc. This will potentially increase crash frequency. Bonneson and Zimmerman (2004a) reported a significant increase in crash frequency with the increase of operating speed. In addition, higher driving speed is, in general, likely to result in more severe injuries in crash. Therefore, increasing speed may generate more revenue from RLC, but is also likely to diminish safety. 3.4. Increase V/C ratio through signal re-timing Fig. 3. Probability of stopping as a function of travel time to stop line in different speeds (Bonneson et al., 1994). where S1 is stopping distance; S2 is travel distance during yellow; Y is yellow duration. Assuming a 4 s yellow interval and using other parameter values in Eq. (1) as recommended by ITE, dilemma zones at different driving speeds is shown in Fig. 2. When driver's speed is above a threshold value (e.g. about 41 mph (66.0 km/h) in the Fig. 2 case), they will be in a dilemma zone. At higher speeds, the range of dilemma zone increases. Increasing drivers’ speed will increase the probability of catching them in a dilemma zone and thus the probability of RLR. Many studies have indicated that, controlling for travel time to the intersection, highspeed drivers are less likely to stop than low-speed drivers (Allsop et al., 1991; Panagiotis, 2007; Verghese and Alex, 2010). Fig. 3 shows the relationship between the probability of stopping and the time to stop line in different speeds from Bonneson et al. (1994). Bonneson et al. (2002) reported a 45% increase in the redlight violation frequency when the running speed was increased by 10 mph (16.1 km/h) and a 33% decrease in the violation frequency when the running speed was decreased by 10 mph (16.1 km/h). To increase revenue from RLC, raising the speed limit may be the most straightforward and effective way of increasing driving speed. NCHRP (2003) indicated that operating speed (85th percentile speed) typically exceeded posted speed limit. Even the 50th percentile operating speed was found to either be near or exceed the posted speed limit. The difference between operating speed and posted speed limit varies in different driving environments. Another important factor contributing to drivers' speeding behavior is roadway geometric conditions, e.g. number of lanes, curvature, Drivers are more likely to run a red-light in high volume-tocapacity (V/C) conditions because they either experience more frustration in congested flow conditions or they follow the behavior of the leading vehicle and try to stay in a platoon. Bonneson and Zimmerman (2004a) reported that a decrease in V/C ratio was associated with a decrease in red-light violations. The V/C ratio in the range of 0.6 to 0.7 yielded the lowest number of violations regardless of speed, path length, yellow duration, heavy-vehicle percentage, cycle length, phase duration, or traffic volume. Several other studies also identified correlations between traffic flow rates and the incidence of RLR events (Baguley, 1988; Hasim, 2009; Porter and England, 2000) and crashes (Mohamedshah et al., 2000). Also, many studies indicated that drivers were more likely to run red-light in close-following driving situations (Allsop et al., 1991; Green, 2003; Zhang et al., 2008). Fig. 4 shows the probability of stopping in different car-following situations based on Allsop et al. (1991). Based on these studies, increasing the V/C ratio by timing traffic signal could potentially increase the number of red-light violations. The timing of signal is not likely to change the total traffic volume (V). However, the capacity of intersections (C) is largely impacted by signal timing. Some signal timing measures, like increasing red-togreen ratio, installing an unwarranted dedicated turn signal phase, dedicating longer pedestrian phases, etc., can be considered. Depending on conditions of specific intersections, different signal timing strategies can be designed. RLCs can also be placed on intersections with high V/C ratios, regardless of crash history. Clearly, increasing V/ C ratio by intentionally decreasing the capacity lowers the efficiency of intersection operation, although it may be able to generate more revenue from RLC. 4. Discussion One of the major challenges with implementing RLC policy is the conflict of matching incentives of tangible revenue for industry Q. Yang et al. / Transport Policy 29 (2013) 192–198 Fig. 4. Probability of stopping as a function of travel time to stop line in different car-following situations (Bonneson et al., 2001). Table 2 A list of measures and their effectiveness, safety impacts, and efficiency impacts. Measures Shorten yellow and/or lengthen all-red Shorten cycle length Increase speed limit Increase V/C ratio Effectiveness at Increasing RLR Negative Impact Safety Efficiency Public Support High High Low Low Low High High High Low High High High Demand Dependent Low Low Low Low High Low and the municipality contrasted with external cost savings such as safety and congestion whose value is not easily captured or internalized by the public sector. As such, there is a temptation to modify the signal systems to increase revenue, particularly if restrictive legislation forces municipalities into contractual challenges with vendors (e.g. revenue guarantee requirements). This paper is not advocating compromising safety or operations for revenue generation, but rather highlights a few easy-toimplement engineering tricks to increase RLR. This paper is meant to clarify the debate and highlight motivations for signal phase changes. The public sector can view this and reflect on the motivation for changing signal operations. Public stakeholders can use this paper to identify if their municipality is changing signal timing based on solid engineering evidence or based on increasing revenue. Based on previous discussions, Table 2 summarizes these measures with a qualitative estimation of their effectiveness on increasing red-light violations and thus revenue along with their impacts on safety and efficiency. Since all these measures increase the chance of RLR, they all imply a safety concern. The safety impacts listed in Table 2 indicate the direction and strength of direct consequences of implementing one of the measures. Shortening yellow duration forces drivers to change their yellow light expectations against their habitual decision, which may lead to abrupt braking or indecisive accelerating. These unexpected actions may cause rear-end and rightangle collisions and thus are marked as high safety concern in Table 2. Lengthening all-red, on contrary, is a measure to protect drivers from right-angle collision, though it should not affect short-term RLR behavior. It is marked as low safety concern. Shortening cycle length and increasing V/C ratio are both indirect 197 ways to stimulate drivers’ motives to run a red-light, but less likely to have immediate, large-scale effects on crash frequency. Finally, increasing speed limit (and thus driving speed) increases dilemma situation and indecision to drivers as discussed above. When a collision occurs, high driving speed also implies a higher severity. The last column in Table 2 shows the negative impact on public support. The public support is tied to the effectiveness of the measure on trapping RLR. Citizens do not like the idea of RLC largely because they do not want to be cited when running a redlight. When an engineering measure shows an obvious hint to drivers that this may increase their chance of RLR at RLC equipped intersections, they are more likely to oppose it, especially when they perceive that traffic engineers may have done this intentionally. Therefore, modifying cycle length, yellow duration, and all-red length may have high negative impact on public support to RLC programs since there are clear linkages between those strategies and RLR. In comparison, increasing speed limit and change V/C ratio are implicit to drivers and thus may have lower impact on public support. The effectiveness and impacts of these measures are discussed in the instance that only one single measure is implemented. While each measure has its merits and faults, a combination of more than one measure may produce good results in maximizing RLC revenue and low impacts on safety and efficiency. Since driver behavior varies in different driving conditions and driver populations, the effectiveness of these measures also varies with time and location. Local studies can confirm the effectiveness of such measures. Our study focused on some engineering measures, particularly focusing on low cost modifications of signal phase. Our study is not exhaustive in terms of clever strategies possible to increase RLR. In the survey of literature, the authors noticed that there were some other measures adopted by some transportation agencies for increasing revenue, such as removing “Signal Ahead” signs. We also do not consider effects of changing signal head configuration and intersection geometry, which may potentially change driver's RLR behavior. For example, in areas with abundant heavy vehicles, signal head placement to mitigate occlusion from following vehicles could influence RLR. A few other measures like actuated signal control and countdown timer may affect RLR behavior as well, as reported by Long et al. (2011, 2013). But since their effects are inconclusive and demand further research, they are not included in this paper. Acknowledgments The research is supported by funding from the Civil and Environmental Engineering Department and Southeastern Transportation Center at the University of Tennessee. References Allsop, R.E., Brown, I.D., Groeger, J.A., Robertson, S.A., 1991. Approaches to modelling driver behaviour at actual and simulated traffic signals, Transport and Road Research Laboratory Contractor Report 264. Crowthorne, U.K. p. 77. Awadallah, F., 2009. A legal approach to reduce red light running crashes. Transportation Research Record: Journal of the Transportation Research Board 2096, 102–107. Baguley, C.J., 1988. 'Running the red’ at signals on high-speed roads. Traffic Engineering and Control 29, 415–420. Bonneson, J., Brewer, M., Zimmerman, K., 2001. Review and evaluation of factors that affect the frequency of red-light running. Texas Department of Transportation Federal Highway Administration. Bonneson, J., Zimmerman, K., 2004a. Development of guidelines for identifying and treating locations with a red-light running problem, Texas Department of Transportation and Federal Highway Administration, Report FHWA/TX-05/04196-2. 198 Q. Yang et al. / Transport Policy 29 (2013) 192–198 Bonneson, J., Zimmerman, K., Brewer, M., 2002. Engineering countermeasures to reduce red-light-running. FHWA/TX-03/4027-2. Texas Department of Transportation and Federal Highway Administration. Bonneson, J.A., McCoy, P.T., Moen, B.A., 1994. Traffic Detector Design and Evaluation Guidelines. Nebraska Dept. of Roads, Lincoln, Nebraska Report no. TRP-02-31-93. Bonneson, J.A., Zimmerman, K.H., 2004b. Effect of yellow-interval timing on redlight-violation frequency at urban intersections. In: Proceedings of the Transportation Research Board 83rd Annual Meeting, Washington, D.C. Brewer, B., Jacobs, D., 2011. Redflex Sues Farragut Over New Traffic Camera Law. 〈http://www.knoxnews.com/news/2011/nov/17/redflex-sues-farragut-over-newtraffic-camera/〉. Burkey, M.L., Obeng, K., 2004. A detailed investigation of crash risk reduction resulting from red light cameras in small urban areas. Urban Transit Insititute, NC A&T State University Report no. DTRS93-G-0018. Cunningham, C.M., 2004. Evaluating the Use of Red Light Running Photographic Enforcement Using Collisions and Red Light Running Violations. North Carolina State University, Raleigh, NC. (Master's Thesis). Chatterjee, A., Cate, M.A., 2009. Issues and Impact of Red Light Camera and Automated Speed Enforcement. Southeastern Transportation Center, Knoxville, TN. Fitzsimmons, E.J., Hallmark, S.L., Orellana, M., McDonald, T., Matulac, D., 2009. Investigation of violation reduction at intersection approaches with automated red light running enforcement cameras in clive, iowa, using a cross-sectional analysis. Journal of Transportation Engineering 135, 984–989. Fleck, J., Smith, B., 1999. Can we make red-light runners stop?: red-light photo enforcement in San Francisco, California. Transportation Research Record: Journal of the Transportation Research Board 1693, 46–49. Florida Statute Chapter 316. State Uniform Traffic Control. Garber, N.J., Gadiraju, R., 1989. Factors affecting speed variance and its influence on accidents. Transportation Research Record 1213, 64–71. Garber, N.J., Miller, J.S., Abel, R.E., Eslambolchi, S., Korukonda, S.K., 2007. The impact of red light cameras (Photo-Red Enforcement) on crashes in Virginia, Federal Highway Administration and Virginia Department of Transportation Report FHWA/VTRC 07-R2. Garber, N.J., Miller, J.S., Eslambolchi, S., Khandelwal, R., Mattingly, K.M., Sprinkle, K. M., Wachendorf, P.L., 2005. An Evaluation of Red Light Camera (Photo-Red) Enforcement Programs in Virginia. Final Report (VTRC 05-R21). Virginia Transportation Research Council. Gårder, P., 2004. Traffic Signal Safety Analysis of Red-Light Running in Maine—Final Report. Georgia General Assembly, 2008. House Bill 77. Green, F.K., 2003. Red light running. ARRB Transport Research, Vermont South, Victoria, Australia. Research Report ARR 356. Han, L.D., Li, J., 2007. Short or long … which is better? Probabilistic approach towards cycle length optimization. Transportation Research Record: Journal of the Transportation Research Board 2035, 150–157. Hasim, M.B., 2009. Effect of various types of traffic signal on red light running. Universiti Teknologi Malaysia, Skudai, Johor, Malaysia. Huang, H., Chin, H., Heng, A., 2006. Effect of red light cameras on accident risk at intersections. Transportation Research Record: Journal of the Transportation Research Board 1969, 18–26. IIHS, 2010. Summary of Decisions Concerning Automated Enforcement. 〈http:// www.iihs.org/laws/auto_enforce_cases.html〉. ITE, 1989. Determining vehicle change intervals—a proposed recommended practice. ITE Journal—Institute of Transportation Engineers 57. KCBD-TV, 2007. Lubbock City Council Delays Red Light Cameras. 〈http://www.kcbd. com/Global/story.asp?S=6129121&nav=3w6y〉. KDFW-TV, 2007. Investigation: Red Light Camera Red Alert. Kent, S., Corben, B., Fildes, B., Dyte, D., 1995. Red Light Running Behavior at Red Light Camera and Control Intersections. Report Prepared for VicRoads by Monash University Accident Research Centre. Lazenby, B., 2008. Quick Light Leads to Refunds for 176 Drivers. 〈http://www. timesfreepress.com/news/2008/mar/13/quick-light-leads-refunds-176-drivers/〉. Long, K.J., Han, L.D., Yang, Q., 2011. Effects of countdown timer on driver behavior after yellow onset at chinese intersections. Traffic Injury Prevention 12 (5), 538–544. Long, K.J., Liu, Y., Han, L.D., 2013. Impact of countdown timer on driving maneuvers after the yellow onset at signalized intersections: an empirical study in Changsha, China. Safety Science 54, 8–16. Lum, K.M., Wong, Y.D., 2003a. A before-and-after study of driver stopping propensity at red light camera intersections. Accident Analysis and Prevention 35, 111–120. Lum, K.M., Wong, Y.D., 2003b. A before-and-after study on red-light camera installation. ITE Journal-Institute of Transportation Engineers 73, 28–32. Lum, K.M., Wong, Y.D., 2003c. Impacts of red light camera on violation characteristics. Journal of Transportation Engineering 129, 648–656. Maccubbin, R.P., Staples, B.L., Salwin, A.E., 2001. Automated Enforcement of Traffic Signals: A Literature Review. Federal Highway Administration, Washington, DC. Martinez, A., 2004. Error Slams Traffic Tickets Into Reverse. Los Angeles Times Los Angeles, CA. McGee, H.W., 2003. Making Intersections Safer: A Toolbox of Engineering Countermeasures to Reduce Red-Light Running. Institute of Transportation Engineers, Washington, DC. Mohamedshah, Y.M., Chen, L.W., Council, F.M., 2000. Association of Selected Intersection Factors With Red-Light-Running Crashes. Federal Highway Administration Highway Safety Information System Summary Report. Moran, C., 2012. City Settles Suit. Cameras Down in 2 Months. Houston Chronicle. 〈http://www.chron.com/news/houston-texas/article/City-settles-suit-camerasdown-in-2-months-3160980.php〉. Naples News Daily, 2010. Judge Rules Red-Light Cameras Unconstitutional in East Coast City. 〈http://www.naplesnews.com/news/2010/feb/22/judge-red-light-ca meras-unconstitutional-aventura/〉. NCHRP, 2003. Design Speed, Operating Speed, and Posted Speed Practices. Transportation Research Board NCHRP report 504. Olson, B., 2010. Houston Voters Reject Red-Light Cameras By Wide Margin. Houston Chronicle. 〈http://www.chron.com/cars/article/Houston-voters-reject-red-lightcameras-by-wide-1619320.php〉. Ovalle, D., 2012. Florida Supreme Court to Hear Red-Light Camera Issue. TheMiamiHerald. 〈http://www.miamiherald.com/2012/11/06/3084981/florida-supre me-court-to-hear.html#storylink=cpy〉. Panagiotis, P., 2007. Driver behaviour, dilemma zone and safety effects at urban signalised intersections in Greece. Accident Analysis and Prevention 39, 147–158. Persaud, B., Council, F., Lyon, C., Eccles, K., Griffith, M., 2005. Multijurisdictional safety evaluation of red light cameras. Transportation Research Record: Journal of the Transportation Research Board 1922, 29–37. Pinkerton, J., 2010. Houston will Turn off Red-Light Cameras on Monday morning. Houston Chronicle. 〈http://www.chron.com/news/houston-texas/article/Hous ton-will-turn-off-red-light-cameras-on-Monday-1609250.php〉. Porter, B.E., England, K.J., 2000. Predicting red-light running behavior: a traffic safety study in three urban settings. Journal of Safety Research 31, 1–8. Radalj, T., 2001. Evaluation of Effectiveness of Red Light Camera Programme in Perth. Main Roads Western Australia. Retting, R., Greene, M., 1997. Influence of traffic signal timing on red-light running and potential vehicle conflicts at urban intersections. Transportation Research Record: Journal of the Transportation Research Board 1595, 1–7. Retting, R.A., Chapline, J.F., Williams, A.F., 2002. Changes in crash risk following re-timing of traffic signal change intervals. Accident Analysis and Prevention 34, 215–220. Retting, R.A., Ferguson, S.A., Farmer, C.M., 2008. Reducing red light running through longer yellow signal timing and red light camera enforcement: results of a field investigation. Accident Analysis and Prevention 40, 327–333. Retting, R.A., Williams, A.F., Farmer, C.M., Feldman, A.F., 1999a. Evaluation of red light camera enforcement in fairfax, Va., USA. ITE Journal (Institute of Transportation Engineers) 69, 30–34. Retting, R.A., Williams, A.F., Farmer, C.M., Feldman, A.F., 1999b. Evaluation of red light camera enforcement in Oxnard, California. Accident Analysis and Prevention 31, 169–174. Roper, B.A., Fricker, J.D., Kumares, C.S., Montogomery, R.E., 1990. The effects of the All-Red Clearance Interval on Intersection Accident Rates in Indiana. Purdue University. Indiana Department of Transportation Joint Highway Research Project. FHWA/IN/JHRP-90/7. Ruby, D.E., Hobeika, A.G., 2003. Assessment of red light running cameras in Fairfax County, Virginia. Transportation Research Board 2003 Annual Meeting CDROM, Washington D.C. Seyfried, R.K., 2004. Engineering Intersections to Reduce Red-Light Running. Northwestern University Center for Public Safety, Evanston, IL. Shatzman, B., 2005. Traffic Signal Snafu Has City Seeing Red. The Oakland Tribune. Souleyrette, R.R., O’Brien, M.M., McDonald, T., Preston, H., Storm, R., 2004. Effectiveness of All-Red Clearance Interval on Intersection Crashes. Center for Transportation Research and Education. Iowa State University. Springfield News-Leader, 2007. Camera Spots 11 Red-Light Runners. 〈http://www. news-leader.com/apps/pbcs.dll/article?AID=/20070517/NEWS01/705170382/1007〉. Stein, H.S., 1986. Traffic signal change intervals: policies, practices, and safety. Transportation Quarterly 40, 433–445. Sun, X., Han, L.D., Urbanik, T., 2011. Secondary coordination at closely-spaced actuated traffic signals. ASCE Journal of Transportation Engineering 137 (11), 751–759. theNewspaper.com, 2009. Florida: Lawsuits Challenge Red Light Camera Legality. 〈http://www.thenewspaper.com/news/28/2881.asp〉, vol. 2012. the Newspaper.com, 2012. Tennessee General Assembly, 2008. House Bill 3854. Tracy, Southerland, Beavers, Summerville, Johnson, Ketron, McNally, Bell, Campfield, 2011. State of Tennessee Public Chapter no. 425, Senate Bill no. 1684. Tobia, P.J., 2006. Yellow Light Blues. The Nashville Scene. 〈http://www.nashvilles cene.com/nashville/yellow-light-blues/Content?oid=1193119〉. Van der Horst, R., Wilmink, A., 1986. Drivers' decision-making at signalized intersections: an optimization of the yellow timing. Traffic Engineering and Control 27, 615–622. Verghese, V., Alex, S., 2010. Modelling Driver Behaviour at the Onset of Yellow Phase at Signalized Intersection. ICTT Civil Engineering Papers. Washington, S., Shin, K., 2005. The impact of red light cameras (automated enforcement) on safety in Arizona, Arizona Department of Transportation, Report FHWA-AZ-05-550. Zador, P., Stein, H., Shapiro, S., Tarnoff, P., 1985. Effect of signal timing on traffic flow and crashes at signalized intersections. Transportation Research Record. Zhang, L., Zhou, K., Zhang, W.-b., Misener, J.A., 2008. Empirical Observations of Red Light Running at Arterial Signalized Intersection. Path Technical Note 2008-1. Institute of Transportation Studies University of California, Berkeley.