1

2 Modelling & Transportation Infrastructure Investment Decision-MakingPresented at: Public Investment for the Long Term: The Case of Public Transportation SPPG Policy Institute, Toronto, January 18, 2016 Eric J. Miller, Ph.D. Professor, Dept. of Civil Engineering Director, UTTRI University of Toronto

3 “All models are wrong; some are useful.”George Box ( ) How useful have / are / could be travel demand models in transportation infrastructure investment decision-making?

4 Modelling & Decision-Making (1)The role of modelling in transportation infrastructure investment & decision-making has been decidedly mixed (to say the least) over the years. Certainly locally we have too many examples of politicians making decisions on the basis of essentially no evidence – more often than not with unwelcome results.

5 Modelling & Decision-Making (2)And it’s not just politicians: BANANA: Build Absolutely Nothing Anywhere Near Anything CAVE: Citizens Against Virtually Everything NOPE: Not on Planet Earth. Rina Cutler, former Philadelphia Deputy Mayor of Transportation People clearly often over-estimate the costs & risks of change and underestimate its potential benefits. And vice versa for the status quo.

6 Modelling & Decision-Making (3)While recognizing the major influence of politics, etc. on decision-making, without strong data and strong model-based evidence how can it be possible escape the dominance of ideology, bias and just plain ignorance? The future is clearly a very uncertain place and the notion of estimating future outcomes is perhaps foolish. And yet, should we just make decisions without any attempt to follow the implications of what we think we know about traveller behaviour & needs, transportation system performance and our “best guesses” at likely future conditions? VKT Time Base Year Forecast Horizon Historical Trend Projection Dynamic, path-dependent response to policy initiatives Static equilibrium projection

7 Three Questions: ILUTE Model Output Are cities and their transportation systems “knowable” / “model-able” in ways that can be useful for policy analysis & decision support? If yes, do we actually have the tools we need? If yes, how can these models be most effectively used in policy analysis & decision support?

8 Clearly cities and their infrastructure systems are exceedingly complex: they are systems of systems. Demographics Land Use Economy Transportation + - Accessibility Mobility Productivity Congestion Pollution/GHG Accidents Loss of Land + - QUALITY OF LIFE

9 TRANSPORTATION SYSTEMThey are also social, economic, cultural and political systems, not just technological in nature. INPUTS URBAN ACTIVITY SYSTEM TRANSPORTATION SYSTEM Land Development Transportation Network Demographics Location Choice Automobile Ownership Regional Economics Activity Schedules Travel Demand Government Policies Activity Patterns Network Flows

10 Cities & Complexity: Jane JacobsJane Jacobs in Death & Life of Great American Cities (1961) famously described “the type of problem a city is” as being “a problem in organized complexity”. This was particularly prescient in that systems theory was in its infancy and complexity theory did not yet really exist.

11 Rittel & Webber: Wicked ProblemsA decade-plus after Jacobs, Rittel & Webber (1973) described planning problems as “wicked”, as opposed to the “tame” problems that the physical sciences are designed to address: Goals are difficult (impossible) to agree upon “Optimization” is impossible Problem definition itself is problematic “Problems” are never “solved”, we (at best) iteratively “re-solve” them over & over

12 Wicked Problems, cont’dRittel & Webber represent typical of reaction against “systems analysis” as it was practiced in the 1960’s during the first generation of high-speed computing and an associated belief that operations research and systems theory could “solve” major policies and even win wars (David Halberstam, The Best and the Brightest, 1972). Generally, reality fell far short of these claims (I.R. Hoos, Systems Analysis in Public Policy: A Critique, University of California Press, 1972).

13 40 Years Later … Cities remain wicked problems in organized complexity. But does that mean that they cannot be systematically studied & understood? “Optimal” “solutions” may be difficult or impossible to achieve but can we not bring analysis-based evidence to bear on decision-making?

14 Modelling Cities (1) Bettencourt (2013):“There is no longer much of an excuse to ignore many of the measurable properties of cities. Cities across the globe and through time are now knowable like never before, across many of their dimensions: social, economic, infrastructural and spatial.”

15 Modelling Cities (2) Our theory, data, models & computing power have grown tremendously. If we avoid the hubris of the 1960’s (that everything is quantifiable; that everything is optimizable; that everything is under our control) can we not utilize the methods, data and understanding we have developed since then to address significant urban issues? And what is the alternative if we don’t?

16 Travel Demand ModellingIII II Key System Elements T – transport system A – activity system F – flows & transport system performance System Interactions/Feedbacks I Market demand-supply interactions determine flows & system performance II System performance (accessibility) influences activity system markets III Gov’t, public & private service providers respond system demand & performance Source: Manheim, M.L. (1978) Fundamentals of Transportation Systems Analysis Volume 1: Basic Concepts, MIT Press A F Travel Demand Modelling Standardized approaches for modelling travel demand have been in use since the dawn of digital computing in the 1950’s. While often useful, these methods have been deeply criticized for decades and certainly have often been misused & abused. Population & Employment Forecasts Trip Generation Trip Distribution Mode Split Trip Assignment Transportation Network & Service Attributes Link & O-D Flows, Times, Costs, Etc. Source: Meyer, M.D. & E.J. Miller (2001) Urban Transportation Planning, New York: McGraw-Hill

17 Emerging Models (1) Fortunately we are in the early days of significantly improved travel demand models that exploit vastly enhanced computing capabilities, data, theory & empirical experience. As I speak we are using an agent-based microsimulation model that simulates “a day in the life” of every person and household in the GTHA to evaluate transit investment options for the City of Toronto.

18 Emerging Methods (2) These persons/households are synthesized (i.e., not actual real people) but they are (hopefully) representative of a future year population in the region. I.e., we are creating a virtual GTHA that “looks like” the real thing but is, obviously (and inherently) a simplified, abstracted model of the real thing. But it allows us to explore a wide variety of “what if” questions and to generate a lot of detailed information that is hopefully useful in: Better understanding travel behaviour. Better understanding the impacts of proposed policy (or policies) on travel behaviour and system performance. Much more in-depth evaluation of benefits & costs, strengths & weaknesses, winners & losers.

19 Agent-Based Modelling (ABM)An intelligent object is an agent. (“an object with attitude” – Paul Waddell). Agents: perceive the world around them make autonomous decisions act into the world ABM provide a an efficient, highly extensible framework for modelling complex human socio-economic activity. Person 1 Agenda Schedule Person 2 Household Dwelling Unit Zone Worker Job Firm Building Vehicle

20 E.g., Daily Tour-Based Trip-Makingtime of day x y home work shop travel between activity locations by transit by auto Base Case 1: Shopping episode on the way home from work. Auto used for the daily activity pattern. A transit improvement causes the person to shift to transit for the journey to/from work. In order to still go shopping, a new home-based auto-drive trip chain is generated. Auto usage & emissions will be under-estimated by a trip-based model.

21 Household-Person ModelsAllocation of resources, assignment of tasks Both households & persons must be simultaneously modelled to “properly” deal with many system components. Household level decisions/processes include: housing location/type choice automobile ownership demographics/household structure/lifecycle stage activity/travel scheduling Households: share resources among household members constrain member behavior condition member decision-making generate activities Requests for resources, availability for tasks Person 1 Person 2 Pers1 Pers 2 Car 1 Time Request for car Allocation of the car to a given person

22 TASHA: Travel/Activity Scheduler for Household AgentsTASHA is the ABM that UofT has built for the GTHA. Originally developed using 1996 travel survey data for the GTHA. The activity scheduler was validated against 2001 survey data. Has been experimentally applied to Montreal, London & Changzhou, China. Has been completely re-estimated/calibrated using 2011 TTS data for operational deployment by the City of Toronto.

23 = “Gap” in Project AgendaTASHA generates the number of activity episodes from a set of “projects” that a person (or household) might engage in during a typical weekday. It also generates the desired start time and duration of each episode. It then builds each person’s daily schedule, adjusting start times and durations to ensure feasibility. Travel episodes are inserted as part of the scheduling process. Activity Episode Frequency, Start Time and Duration Generation (a) Draw activity frequency from marginal PDF (b) Draw activity start time from feasible region in joint PDF (c) Draw activity Duration from feasible region in joint PDF Joint PDF Joint PDF Activity Frequency Start Time PDF Start Time Duration Activity Frequency Feasible Start Times Feasible Durations Scheduling Activity Episodes into a Daily Schedule Work Project Work School Project Other Project Other Shopping Project Shop 1 Shop 2 : : Person Schedule At-home At – Home Work Shop 1 Other Other At-home Shop 2 At-home = “Gap” in Project Agenda = Activity Episode = Travel Episode

24 Tour-Based Mode ChoiceChain c: 1. Home-Work 2. Work-Lunch 3. Lunch-Meeting 4. Meeting-Work 5. Work-Home mN = mode chosen for trip N Drive Option for Chain c Non-drive option for Chain c m1 = drive m5 m4 m3 Sub-Chain s: 2. Work-Lunch 3. Lunch-Meeting 4. Meeting-Work m2 m1 TASHA’s tour-based mode choice model: Handles arbitrarily complex tours and sub-tours. without needing to pre-specify the tours Dynamically determine feasible combinations of modes available to use on tours. Modes can be added without changing the model structure. Cars automatically are used on all trips of a drive tour. Drive for Sub-chain s Non-drive for Sub-chain s m2 = drive m3 = drive m4 = drive m4 m3 m2 m5 = drive

25 Vehicle Allocation & RidesharingSince we are modelling each person within a specific household context, we can explicitly model within-household vehicle allocation and complex ridesharing behaviour.

26 How Should Models Not Be Used?Post-facto justification of political decisions. Barrier to debate & experimentation. Hard-wiring “desired” decisions/outcomes into the analysis. Failure to exploit the richness of information generated by the model. Or, just as bad – models are ignored completely.

27 How Should Models Be Used? (1)Policy analysis vs. forecasting. Too often we focus on modelling as a forecasting exercise to generate “the number”. Models are far better at policy analysis, in which the relative benefits & dis-benefits of alternatives are compared.

28 How Should Models be Used? (2)Models should be used: As a laboratory within which we can experiment with alternative system designs & policies. To explore alternative paths into the future and their likely consequences. To learn & educate. To broaden the debate. To understand the nature & importance of uncertainty. Be “another voice at the table”.

29 How Should Models Be Used? (3)ENVIRONMENT S U T A I N B L E R P O ECONOMY SOCIETY NEIGHBOUR - HOODS GOVERNANCE INFRA STRUCTURE FINANCING “The Four Pillars of Sustainable Urban Transportation”, Kennedy, et. al (2005) We are potentially entering an era in which advanced models can provide a much richer, flexible and powerful platform for policy analysis & decision support. Whether we exploit this capability depends on whether planners, politicians & the public actually want: Better information. Informed debate. Flexibility in desired outcomes.

30 Acknowledgements to the UofT modelling team (past & present):Matt Austin Ahad Bekaei Juan Antonio Carrasco Leo Chen Franco Chingcuanco Len Eberhard Ilan Elgar Bilal Farooq Yiling Deng Leila Dianet Sean Doherty Jared Duivestein Wenli Gao Martin Giroux-Cook Kathryn Grond Ahsan Habib Khandker Habib Torsten Hahmann Murtaza Haider Michael Hain Ayad Hammadi Jiang Hao Tony Harapin Chris Harding Adam Harmon He He Marianne Hatzopoulou Brian Hollingworth Nik Krameric David King Peter Kucirek Marek Litwin Wenzhu Liu Greg Lue Kouros Mohammadian David McElroy Monika Nasterska Trajce Nokolov Gurbani Paintal Winnie Poon David Pritchard Anna Pushkar Matt Roorda Adam Rosenfield Paul Salvini Bruno Santos Fernanda Soares James Vaughan David Wang Joshua Wang Marcus Williams Yunfei Zhang THANK YOU! QUESTIONS?

31

32 Achieving efficient public decisions in the transport sectorJonas Eliasson Professor Transport Systems Analysis, KTH Department for Transport Science & Centre for Transport Studies

33 The benefits of formal cost-benefit analysisWhat does experience in metropolitan areas internationally tell us about the strengths and weaknesses of cost‐benefit analysis or other decision rules in assessing the short‐ and long‐term factors to be considered in decision‐making about urban transportation?

34 Benefit/cost ratios of 500 investment candidatesTop 150 CBA has many weaknesses – many uncertainties Valuations of effects Scenario assumptions The ”good” are much better than the ”bad” – despite being shortlisted by professionals!

35 The ranking is not very sensitive to assumptionsChanges in Top 150 Double freight benefits 14 Double safety benefits 22 Double emission benefits 5 Double person travel benefits 11 Differentiated VoT’s Double oil price 2 No plugin hybrids 1 Trend break in car ownership Strong carbon policy package 3

36 … but cost/benefit efficiency has limited effectKanske vi vill ha denna??? Government Transport Administration Left outside

37 Implementing efficient but unpopular policiesWhat do we know about how the framing and communication of issues affects public attitudes toward various urban transportation options? Is there a bias toward the status quo? How can public trust be built, in a context of uncertainty about the projected costs and benefits of urban transportation options?

38 Congestion pricing: It works. Stable decrease 20% across cordonYes it works. These are figures from Stockholm

39 What 20% less traffic does to congestion

40 Difficult to predict or remember behaviourAccording to surveys before, traffic would decrese ~5% According to surveys after, traffic had decreased ~5% The real figure was 30% ~30% switched from negative to positive during the first year A year later, half of those claimed that they had had the same opinion all along

41 The U-curve of support Support Referendum Decision Charges introducedGovt. decision That it worked was perhaps no surprise The big surprise was the swing in public support. Remember that CC makes the generalised cost of driving go UP for most people. What caused this swing? Typical pattern for CC and other policies

42 So why is congestion pricing so uncommon?Many win a little, a few lose a lot Distribution of political power, costs, benefits Strong public emotions about congestion – but congestion pricing not perceived as solution Few have strong emotions for ”efficiency”; hence little political upside

43 Determinants of support for congestion pricingSelf-interest: Personal costs and benefits Public welfare, especially enviromental effects Trust in government Attitude to pricing as a ”fair” allocation instrument

44 Support for CC split by how much people payNo car Have car, never pay Pay sometimes Pay often

45 Drivers of attitude changeIt’s better than you thought It’s not as bad as you thought Reframing – viewing it as something else Status quo bias

46 Questions – answers What do we know about how the framing and communication of issues affects public attitudes toward various urban transportation options? Rationality is not enough to get the political ball rolling: positive emotions necessary to make people care Build alliances: experts + emotions + solving political risks Is there a bias toward the status quo? Yes – a lot How can public trust be built, in a context of uncertainty about the projected costs and benefits of urban transportation options? Be honest about uncertainties. Blame the experts. Use trials. Be prepared to change your mind. Get an institutional setting to survive the valley of political death

47

48 Attitude formation (long run)New attitudes formed by associating to attitudes to ”similar” issues What is ”similar” depends on framing New attitudes are less stable – can be re-associated … especially if re-framed Politics often a battle of framing which existing attitudes and values should a new issue associate to Gaining political ground often requires re-framing of issues

49 Framing congestion charges in Stockholm: A story in 4 acts”Congestion charges gives efficient resource allocation” Few have strong attitude to ”efficient resource allocation” Little emotion => little political upside Noone cares except some transport economists Eliasson, J. (2014) The role of attitude structures, direct experience and framing for successful congestion pricing. Transportation Research A 67,

50 Act 2 ( ): ”Congestion charges is an environmental measure” Strong emotions => potentially large political upside CC looks similar to other such measures => easy to associate to existing environmental attitudes Enters agenda of Green party, environmental NGOs etc.

51 The battle for moral high ground Act 3 ( ): The battle for moral high ground Opponents try to associate to ”tax”, ”harm the poor”, ”unfair” ”restriction of freedom” Preferred term: ”congestion tax” or ”road toll” Proponents try to associate to ”environment”, ”user pays”, to some extent ”anti-rich” and ”anti-car” Preferred term: ”environmental charge” Results in polarization – e.g. alienation of car drivers The less affected people are, the less developed are their attitudes, and the more volatile their attitudes are Unaffected car owners decreased their support the most

52 Reframing and emotional discharging Act 4 (2007-): Reframing and emotional discharging After positive referendum, new government keeps CC ... but earmarks revenues to multi-billion bypass motorway tunnel Reframing from ”anti-car” to ”efficiency” and ”revenue source” ”It’s OK to be a car driver, but please drive less in the city in rush hours” Less emotions From moral domain back to technical-rational domain The latter less emotional => less political interest

53