TECHNICAL NEWSLETTER - SCOR
TECHNICALNEWSLETTER#44 - July 2018IMPLICATIONS OF AUTOMATED VEHICLESON THE MOTOR MARKET:the German modelINTRODUCTIONAutomated vehicles (AV) are keenly anticipated for thebenefits they are expected to bring to society: greatersafety, fewer traffic accident victims, improved access tomobility, and more efficient traffic flow resulting in reducedemissions. The increasing use of automated vehicles willchange the insurance landscape permanently, with thelower likelihood of accidents greatly reducing premiumsin motor insurance. With an expectation of widespreaduse of fully autonomous vehicles, those which provide thehighest level of safety, some studies estimate this reductionat more than 70% of the market premium volume by 20501.However, to assess the implications of automated vehicleson a specific market, the time horizon and severity of theultimate impact must be separately estimated. The timehorizon depends on the penetration rate of automatedvehicles in the underlying motor portfolio; the severityimpact for insurers also depends on this penetration rateand on the new risk premiums, which will be lower thanthe risk premiums for conventional vehicles. In additionto this, the different stages of this technological progress– commonly referred to as levels of automation – must alsobe considered, along with their corresponding efficiency.In the earlier stages, technological capability is quite low,only affecting the risk of accidents to a limited extent. Butin the later stages, technological performance is projectedto surpass human limitations and eradicate human error,which is the main cause of road traffic accidents. That beingsaid, nobody really knows if or when such technologicalperfection will actually be reached.This article analyses the penetration of the different levelsof automation and the corresponding change in motorpremiums for a sample market. Germany is chosen as theobserved market due to its mature motor portfolio and itscharacteristics. Because 8% of the existing motor vehicles arecapable of level 2 automation2, the German motor portfoliois already fairly exposed to this emerging risk. In addition,due to market drivers such as Germany’s legislation andthe developing infrastructure, it is assumed that Germany’smotor market will experience a swifter transformation thanthose of other countries.Although the key assumptions described in the followingcan of course be transferred to other markets, the countryspecific characteristics of each market may influence theweighting and importance of these assumptions.1 – KPMG, The Chaotic Middle: The Autonomous Vehicle and Disruption in Automobile Insurance. 2017.2 – DAT, DAT Report 2017. 2017.SCOR P&C - TECHNICAL NEWSLETTER #44 - JULY 20181
AUTOMATED VS. AUTONOMOUS DRIVINGIn 2013 the Society of Automotive Engineers (SAE) proposedlevels of automation to create a common classification systemand to establish a common understanding of the technological progress of automated vehicles3. According to thisdifferentiation, the transformation of a conventional vehicleinto an entirely automated vehicle will take place over sixlevels of automation, in which level 0 represents conventional vehicles and level 5 fully automated, or autonomous,vehicles. In accordance with these levels, the dynamic drivingtask is gradually shifted from the human driver to theautomation system. Although the terms “automated” and“autonomous” are sometimes used interchangeably, theydo not have the same meaning. As proposed by the SAE,the designation depends on the degree of technology inthe vehicle. If there is no or only limited technology used toassist the driver, this is considered as a conventional vehiclewith an automation level of 0-1. If a vehicle depends on anonboard computer that uses automation to determine andimplement driving algorithms, but no artificial, self-learningintelligence nor cloud-based “hive mind” is implemented,such vehicle is deemed to have an automation level of 2-34.By implication, truly autonomous behaviour is thereforeexpected in levels 4 & 5, where the systems are capable offull-time performance of all aspects of the dynamic drivingtask, fully monitoring the driving environment with no needDID YOU KNOW?Since 2001, having the same understandingand definition of a risk has become a crucial issuefor the insurance sector. For automated vehicles,the literature mainly differentiates between automationlevels according to the SAE. However, legislators do notnecessarily follow this path: in the United Kingdom,the Vehicle and Technology Aviation Bill does notclearly distinguish between the automation levels.This leaves the bill ambiguous in terms of how to apply itto the respective levels of automation. In Germany,the legislator has even chosen another designation forthe respective automation levels (e.g. level 4 is referredas “full automation” which corresponds to level 5according to the SAE). These approaches providepotential for misunderstandings, because foreignparties may not understand the legal situation correctlyin the respective countries.for human interaction. The current status quo is that level 2automated vehicles are available and Audi will introducelevel 3 automation within the new A8 this year6. Level 4automation currently only exists within prototypes, whichare tested under strict observation and are not yet publiclyavailable. In this paper, the six levels of automation aredivided into three classes (levels 0 & 1, 2 & 3 and 4 & 5).012345The driver constantlyperforms all aspects ofthe dynamic driving task.No systems intervene only those that warnthe driver.The system can takeover eitheir steeringor acceleration /deceleration. The drivermust continuously carryout the other.The system can takesover both steeringand acceleration /decelerationin a defined use case.The system can takesover both steeringand acceleration /deceleration in a defineduse case. It is capable ofrecognizing its limits ofnotifying this driver.The driver can hand overthe entire driving taskto the systemin a defined use case.The system can takeover the entire dynamicdriving task in all usecases.The driver mustconstantly monitorthe drive.The driver mustconstantly monitorthe drive. He must beready to resume fullcontrol immediately.The driver mustconstantly monitorthe drive. He must beready to resume controlimmediately.The driver mustconstantly monitorthe drive, but be readyto resume control withina given time frame ifthe system so requests.The driver would not berequired at all duringthese cases - neither formonitoring, nor asbackup.The driver is no longerrequired at all.FIGURE 1: LEVELS OF AUTOMATION ACCORDING TO SAESource: 2025AD53 – SAE, Definitions for Terms Related to On-Road Motor Vehicle Automated Driving Systems-J3016. Society of Automotive Engineers: On-Road Automated Vehicle Standards Committee; SAE Pub.Inc., Warrendale, PA, USA, 2013.4 – The Insurance Institute of Canada, Automated Vehicles, Implications for the Insurance Industry in Canada. 2016.5 – 2025AD. Definition: Levels of AD. 2015. Accessed: 16.12.2016; Available from: tion/6 – Intel, Intel Inside New Audi Autonomous Car System. 2017.2SCOR P&C - TECHNICAL NEWSLETTER #44 - JULY 2018
PENETRATION OF AUTOMATED VEHICLES IN GERMANYIt is difficult to determine the time horizon in which fullyautomated vehicles will ultimately impact the underlyingmarket. Due to the uncertainties surrounding automatedvehicles, determining a certain and discrete scenario forthe penetration of automated vehicles is impossible. Toovercome this issue, the concept of a “range of futures”is used, which assumes that each possible scenario can becaptured within two boundary values7. For the penetrationof automated vehicles, these boundary values are derivedfrom the historical penetration of anti-blocking systems(ABS) and electronic stability programs (ESP) in the Germanmotor portfolio. The penetration of ABS was slow, and ittook 20 years until ca. 40% of the underlying motor portfoliowas penetrated. Conversely, ESP penetrated roughly 80%of the motor portfolio in the same time horizon and istherefore a suitable example of rapid penetration. Hence,the penetration of ABS represents the minimum and thepenetration of ESP the maximum boundary value. For theexisting automated motor portfolio in Germany, whichrepresents roughly 8% of the overall vehicles, the penetration of automation levels 2 & 3 is forecast on the basis ofthese boundary values8. Moreover, a delayed introductionin the respective vehicle class segments is also expected andconsidered in the model: when first put into production,level 2 automation was mostly available for high luxurycars; now it is available for mid-range cars too. It is assumedthat the next levels of automation will follow, with a timehorizon based on historical reference values.DID YOU KNOW?The time horizon and severity impact of automatedvehicles is influenced by the specific characteristics ofeach country. In addition to market acceptance drivenby a clear legal framework, the lack of required digitalinfrastructure is a major constraint. Nevertheless, somecountries like the Netherlands, Singapore, the US,Sweden and the UK (partly) meet the requirements formainstream adaption of automated vehicles. Therefore,it is assumed that these countries will be among the firstto gain relevant experience with the transformationcaused by the automation of vehicles.Therefore, it is presumed that the technology will beonly available for high luxury cars in the beginning, formid-range cars after 4 years and for compact models after 7years. This delayed introduction is driven by the costs of thetechnology involved, which are usually significantly higherwhen a product is first put on the market in comparison toa later stage. As an example, the price of the first built-inGPS navigation system in production vehicles was USD 2,000in 1995. Nowadays, these costs are roughly one-tenth of theindexed original price. By implication, the acceptance andavailability of automated vehicles will increase over time asthe technology becomes cheaper, making sensors, electroniccontrol units, etc. affordable and creating a preconditionfor mainstream adaption.DID YOU KNOW?A clear legal framework is a key driver of the acceptance and penetration ofnew technology. Surveys show that for the majority of participants the issues ofliability and guilt must be defined by law before purchasing an automated vehicle9.In Germany, automated vehicles up to automation level 4 are properly governed bythe law. The 8th law amending the Road Traffic Act prescribes that the driver is stillliable in terms of liability for presumed fault and the vehicle owner/ keeper remainsliable in the sense of strict liability10.7 – Proff, H., Entscheidungen beim Übergang in die Elektromobilität: Technische und betriebswirtschaftliche Aspekte. 2015: Springer-Verlag.8 – DAT, DAT Report 2017. 2017.9 – ADAC. ADAC-Umfrage „Autonomes Fahren“. 2016 Accessed: 15.05.2017; Available from: https://www.adac.de/ 20adac.de 281295.pdf.10 – Bundestag. Achtes Gesetz zur Änderung des Straßenverkehrsgesetzes. 2017 Accessed: 20.06.2017; Available from: https://www.bgbl.de/xaver/bgbl/start.xav?start %2F%2F*%5B%40attrid%3D%27bgbl117s1648.pdf%27%5D# bgbl %2F%2F*%5B%40attr id%3D%27bgbl117s1648.pdf%27%5D 1516351777807SCOR P&C - TECHNICAL NEWSLETTER #44 - JULY 20183
Automation levels 4 & 5 are characterized by a high degreeof uncertainty in terms of both the market and technological drivers; besides the unclear jurisprudence, thelack of required digital infrastructure is one of the majorconstraints. According to ABI Research, highly automatedvehicles need at least “5G” internet access, which corresponds to roughly 1.200 MB per second but is not yetavailable11. As reported by the German initiative “5 stepsto 5G”, the 20 largest cities in Germany should have theIn 2018, new test tracks are going into operationin German cities and provincesSyltFlensburgHamburgBremenrequired stable and broad network access by the year202512. Based on historical values of the past ten years, theprivately used vehicles of these cities represent 14.6% ofthe German motor portfolio and 23.6% of newly registeredcars in Germany. Assuming a steady improvement of thenetwork after 2025 in the larger cities only, the penetrationof automation levels 4 & 5 stagnates at a certain point. This isbecause highly automated features cannot work properly insmaller cities and in the countryside, which are characterizedby slow and unstable internet connection. The penetrationof automation levels 4 & 5 in these areas is not expected inthe medium-term, which means that the motor portfolioof the largest cities represents the maximum penetration ofhighly and fully automated vehicles in Germany. Combiningthis assumption with the ones mentioned above, the penetration for automation levels 4 & 5 can be estimated. Incontrast to that, the penetration of automation levels 0 & 1is the complementary set of the penetration of levels 2 & 3and 4 & 5 with the further assumption that newly registeredcars are equipped with the latest technology and that onlylevels 0 & 1-vehicles are withdrawn from the German motorportfolio.BerlinWuppertalHannover - BraunschweigWolfsburg - SalzgitterDID YOU tMerzig - SaarbrückenMetz - Luxembourg Karlsruhe - HeilbonnBad BirnbachWith the advent of automated vehicles, the reliabilityof testing is becoming increasingly important forautomated vehicles. Accidents caused by the vehicles,such as the widely reported Uber accident, couldtemper the acceptance of this technology. In Germany,automated vehicles are thoroughly researchedand tested13.MunichOpening in 2018Opening data unknowIn operationFIGURE 2: NEW TEST TRACKS FOR AUTOMATED DRIVINGSource: GDV.DE, 201711 – ABIresearch. Role of 5G in Automotive and Transportation. 2016 Accessed: 15.05.2017; Available from: o/.12 – Federal Ministry of Transport and Digital Infrastructure. 5G Strategy for Germany. 2017 Accessed: 15.08.2017; Available from: trategy-forgermany.pdf? blob publicationFile.13 – GDV. Diese Städte und Regionen werden 2018 zu Teststrecken. 2018. Accessed: 03.04.2018; Available from: d-regionen-werden-2018-zuteststrecken-258744SCOR P&C - TECHNICAL NEWSLETTER #44 - JULY 2018
According to Figure 3 besides, after 2034 the entire Germanmotor portfolio should be capable of at least level 2automation. Hence, the transformation of the Germanmotor market is expected to take place within the next 22years (at the latest). From 2034 onwards, advanced drivingassistance systems are believed to be commonly availableand fully distributed in the portfolio, increasing the overallsafety level of road traffic. However, most automatedvehicles will only have an automation level of 2 or 3. By2040, less than 30% of vehicles are expected to be capableof automation levels 4 & 5. This prognosis is mainly drivenby the assumption that the broad, stable network requiredfor levels 4 & 5 is only available in bigger cities. In rural areas,only vehicles at automation levels 2 & 3 are expected tooperate. However, if the required 5G-network is publiclyavailable sooner, which is anticipated to happen in Swedenfor example, the higher automation levels should penetratemore rapidly. Overall, the penetration of automated vehiclesand the disappearance of conventional vehicles will directlyinfluence the required risk premiums for motor insurance,due to the anticipated reduction in loss frequency.Penetration ratio100%LEVEL 4 - High Automation&LEVEL 5 - Full Automation90%80%70%60%50%LEVEL 2 - Partial Automation&LEVEL 3 - Conditional Automation40%30%20%10%LEVEL 0 - Conventional vehicles&LEVEL 1 - Driver Assistance0%Years20112015Level 0 -1201920232027Range of futuresbetween levels0 - 1 and 2 - 320312035Level 2 - 32039204320472051Range of futuresbetween levels2 - 3 and 4 - 52055Level 4 - 5FIGURE 3: PENETRATION OF AUTOMATED VEHICLESIN THE GERMAN MOTOR PORTFOLIOSource: SCOR; own evaluationCHANGE IN GERMAN MOTOR INSURANCE PREMIUMSTo assess the severity impact of automated vehicles from aninsurer’s point of view, the anticipated premium reductionmust be determined. In this regard, the premium reductionis defined as the difference between the new required riskpremium for automated vehicles and the forecasted riskpremium for conventional vehicles at market level. Thischange in premium depends on the penetration mentionedpreviously and the reduced likelihood of accidents. This“new” loss frequency can be derived from actual mitigatable and avoidable claims, as well as from the degree oftechnological perfection involved. By analysing the accidentcategories from police/statistical records, mitigatable andavoidable accidents like “errors when overtaking” or“speeding” are identified and evaluated. Depending onthe system’s capabilities, the accident frequency is graduallyadjusted according to the different levels of automation. Inaddition, a distinction between the types of coverage andthe influence of automated features on this cover provides abaseline from which to assess future claims costs. For level 2automation, there is no significant impact on the claimscosts for partial cover, and this trend continues even for thehigher levels of automation. The only exceptions are theftand collision with animals: automated vehicles are easier tolocate and retrieve due to their onboard GPS-system, andcollisions with animals can be greatly reduced thanks tosafety features such as emergency braking systems. However,these two exceptions make up 27% of claims costs forpartial cover, but less than 5% of the total claims cost formotor insurance cover in Germany. Conversely, starting atlevel 2, the implemented systems influence risk premiumsfor fully comprehensive cover which is mainly driven bythe cost of collisions with other vehicles. As stated by theNational Highway Traffic Safety Administration in the US,the onboard safety features of the Tesla autopilot decreasedthe accident rate by 40%14. In this context, the combination of lane-guard, lane-change and other driver assistancesystems significantly influences the risk of accidents, therebyreducing claims costs for fully comprehensive cover. Thistrend will continue for higher levels of automation, wherethe dynamic driving task will be optimally executed by thesystem, eradicating human error. There will be a similarimpact on Motor Third-Party Liability (MTPL) cover. Accidents14 – Muoio, D. Tesla is pushing the insurance industry to prepare for massive disruption. 2017. Accessed: 18.11.2017; Available from: ng-cars-arechanging-insurance-industry-2017-5?r US&IR TSCOR P&C - TECHNICAL NEWSLETTER #44 - JULY 20185
will increasingly be mitigated or avoided as the level ofautomation progresses and eliminates the major driver ofMTPL claims. The risk premium will therefore decrease dueto the reduced likelihood of accidents.Nonetheless, this development has counterbalancing effects,the most important being the increase of Casco value due tothe greater complexity of the cars. As stated in a study byLiberty Mutual, the total cost of a low speed collision for thesame vehicle model (2014 and 2016 4-door entry-level luxurysedan) increases by 92.3% overall, because the 2014 model isnot equipped with the same technology as the 2016 model.Hence, the increase in costs is mainly caused by the damagedparts (i.e. distance sensor and LED headlamps) which are130% more expensive to replace in the 2016 model than inthe 2014 model. Labour costs inflate the claim by a further18%, because more highly trained specialists are needed tocarry out the
4 SCOR P&C - TECHNICAL NEWSLETTER #44 - JULY 2018 SCOR P&C - TECHNICAL NEWSLETTER #44 - JULY 2018 5 Automation levels 4 & 5 are characterized by a high degree of uncertainty in terms of both the market and techno-logical drivers; besides the unclear jurisprudence, the lack of required digital infrastructure is one of the major constraints.
both to further model development and to obtain the full benefits of membership. The SCOR-model is still being developed the latest version of SCOR-model is numbered 7.0. SCOR is a management tool. It is a process reference model for supply-chain management, spanning from the supplier's supplier to the customer's customer. The SCOR-model has been
SCOR is a novel retrieval framework that combines MRF and word2vec to model order and semantics together. SCOR gives state-of-the-art results on source code retrieval task of bug localization. In the process of developing SCOR we also generated semantic word embeddings for 0.5 million software-centric terms from 35000 Java repositories.
The SCOR model is an integrated approach in supply chain management to measure the performance of organizational standards at each level of the stage into the supply chain framework through a benchmark, gap analysis and best practice approach for sustainable development [14]. The supply chain Council (2012) assert that the SCOR-model has
The Supply-Chain Operations Reference (SCOR) model was developed by the Supply-Chain Council (SCC) to assist Þrms in increasing the effectiveness of their supply chains, and to provide a process-based approach to SCM (Stewart, 1997). The SCOR model provides a common process oriented language for communicating among
in their supply chain using the SCOR model will need to extend the model, at least to level four [Calderón, J.L. and Lario, F.C., 2005, p. 4](1) Stadtler meanwhile, et al. (2005) argues that SCOR is a reference model, does not have a mathematical or heuristic method, however it standardizes terminology and
Figure 2.2: Scope of the SCOR-model 7 Figure 2.3: The hierarchy of the SCOR-model 8 Figure 2.4: Linking supply chain management to balanced scorecard 10 Figure 2.5: A supply chain balanced scorecard framework 11 Table of tables Table 2.1: Advantages and disadvantages of the SCOR-model
SUPPLY CHAIN OPERATIONS REFERENCE (SCOR) MODEL OF SUPPLY CHAIN COUNCIL SCOR modeling is referred today as the most popular methodology to analyze the Supply Chain Management (SCM) performance. The SCOR model is developed by Supply Chain Council in 1996 by Pittiglio Rabin Todd & McGrath and AMR Research.
American Revolution in Europe working to negotiate assistance from France, Spain, and the Netherlands. Foreign Assistance French ultimately provided critical military and financial assistance Spain and the Netherlands provided primarily financial assistance to the American cause. A comparison of the resources held by the British and by the colonies: The population of the thirteen colonies .
