It's my pleasure to turn the meeting over to Jocelyn Bauer. Welcome to the national transportation coalition talking operations seminar It Pays to do the Right Thing. I will give a brief introduction before turning over to Darren Buck. Please be advised today's seminar is being recorded, will last approximately an hour and 30 minutes, the first 60 for presentation and the final 30 for audience Question and Answer. If you think of a question during the presentation you can type it into the smaller text box on the left side of your screen, indicate who the question is intended for, unless for all presenters. Type in the syntax and not the white area. The presenters will be unable to answer questions during the presentation, but some of the questions will be answered during the last 30 minutes of the seminar. A file containing the audio and video will be posted to the coalition website within the next week. I will type that address into the chat box shortly. Attendees will be notified of the availability of the session and the closed captioning. We encourage you to ask others to view the presentation. To download a file, click on the file you wish to download and on the box save to computer. I would like to introduce Darren Buck. He's the marketing specialist for the federal highway office of operations and duties include overseeing outreach activities of the national transportation coalition. Prior to joining federal highway, Darren worked with local small business and other federal agencies. I will turn things over to Darren to start us off. Thank you, and thank you all for joining us, the latest in the series of special Webinars we have been doing on various aspects of congestion -- This Webinar will be a little different, today's speakers will about rewards and incentives. One of seven in a series of new primers that discuss the basic rationale, issues surrounding congestion prizing. A link to all seven of these primers can be found on the left side of your screen in the chat box. All recent office of operations publications may be found by going to the website, and clicking on the publications button. Before I introduce you to our first speaker, Dr. Balaji Prabhakar, Department of Electrical engineering and computer science at Stanford University. Recently he's been interested in societal networks, such as transportation, electricity and -- deploying incentive mechanisms to move commuters to off-peak times, a fellow at Stanford University and fellow of the Alfred P. Sloan foundation, award from the National Science Foundation, the [indiscernible] David son prize and the [indiscernible] lectures, corecipient of several Best Paper awards. After that, Nicholas Ramfos, developed and implements various commute alternative programs, policies in the Detroit, Chicago, and San Diego communities. Programs director with the met wash counsel met wash counsel of governments. I will turn it over to Alan Greenberg, to offer a few brief remarks before we begin. Thank you, Darren, and Jocelyn. The strategies that are being focused on today, the incentives and pricing, not of the towing variety, unique, don't get the attention that toll pricing and congestion, tolling normally get, but they are recognized in law, the value pricing pilot program, link for our latest solicitation will be put up. Requesting specifically projects of the non-toll variety, the primer Darren mentioned, I encourage people to look at, as it provides a summary of a lot of different non-toll strategies, what's been done and some of the best researches in those areas. What's excite being the speakers today, they really are the cutting edge of thinking and doing on non-toll pricing. Why the strategies matter, you get similar benefits tolling strategies, but ways that aren't garnering the kind of political opposition you get with toll pricing. In pricing. In Dr. Prabhakar's presentation you will hear about some of the [indiscernible] in India, issue and challenging in the Stanford campus. Nicholas Ramfos is a seasoned hand when it comes to incentives, getting people out of standard drive-alone commute modes. He appreciates very fully the need to offer a range of options, alternatives, and support, but also specifically how the financial incentive, pricing incentive matters in all that. I am slighted to have the two speakers, and I can turn it over to Dr. Prabhakar to begin his talk. Dr. Prabhakar: Thank you, I am delighted to be speaking here. I am going to talk to you about incentive mechanisms. The background in this area of research, certainly colleagues and I at Stanford are engaged in, societal networks. What they are, networks we are familiar with, such as transportation systems, electricity networks, and recycling networks, as well as you can think of water, other utilities. Societal networks, resources society consumes through the aid of using technological mechanisms to deliver resources to us, very critical that human actions be exerted on these networks in such a way that the networks efficiency is pretty high. In this sort of setting, the interesting challenges are how do we design better technology in these -- for delivery of these resources, consumption of these resources and how do we offer effective and simple way of incentivizing the users. Let me focus on the one -- the goal of the presentation, transportation systems. Congestion is something that occurs in any network system, transportation, electricity, computer networks and telephone networks. Since it occurs during the course of operation, it is engineered into these networks to help cope with congestion when -- in carpools, traffic-sensitive lights, tolls, and differential pricing for peak and off-peak use. This is in effect anytime you download or send e-mail, the protocol we use to transfer data. It's also happening in data centers, a lots of standardization work going on. Now, since these methods have been put in place, essentially they address congestion once, once it occurs, helps you cope with it when they kick in. The topic of this talk, on the other hand, incentive mechanisms are all about trying to perhaps postpone or even prevent the occurrence of congestion. We can sort of shift enough of the load away that perhaps it's possible to not face congestion or any congestion at all. How does that work? Well, there are two approaches to doing this. One is the current method, congestion pricing, where you charge drivers who enter a congested zone, as is well-known, commonly used in London, Stockholm, Singapore, a few other places. Our proposal, or what you will see in this talk is to find, create a market approach, let's charge the congesters and pay the decongesters. The presumption is there's enough versed variety in the utility function, some people value time, others money. Those who value time will pay money, those who value money will yield time, remove themselves from the system, collect the money that the system has available. This is one of the main thesis, you will see how a particular experiment we conducted, what happened when we applied this high-level approach. The other sort of technologically speaking, things we do, trying to put intelligence in the vehicles, not on the road, where when we want something to scale, an algorithm, protocol to have the desirable properties scaling to a large number of users, it's wise to distribute and undue burden on the network. Deploying incrementally, the intelligence is built in gradually. The more people there are, the more the intelligence system grows. This is allows us to not worry about the system that should work fully on day one into the entire population. One thing to note, bottom of the slide, that a key aspect of incentive mechanisms, they allow you not to directly address congestion. You may say that's what you are going to do, but it is possible you can give incentives to [indiscernible] to move to off-peak time and if enough commuters do this it will lead to reduction in congestion. You don't have to start on day one saying I will reduce congestion. This is a liberating approach. Two theses we have, we can try to make the "right to congest" into a tradable commodity, and the other piece is that small good deeds do not carry adequate rewards. This applies, for example, the other networks, electricity, recycling, when we turn lights off or the marginal benefit to the system for single light bulb be not burning being small. People tend not to perform them, but collectively, such actions, when they are taken by large numbers of people really improve the efficiency promisingly. tremendously. How do we incent people when small instances don't pay a lot? One option is to pool the individual payments into a pot, and then try to pay out so it [indiscernible] mechanism. In keeping with the game theory, maximization, saying in games with low stakes players are more risk-seeking or certainly risk-neutral. Explaining the reason of [indiscernible] insurance, things like this. Moving to the experiment itself now, the instant project is a synopsis Stanford traffic project. In Bangalore, India, 20,000 people work there, of which 14,000 commute, for participation in the scheme, their [indiscernible] was handled by the Bangalore office. There was launched on October 6 of last year, ran for six months, April 10, a pilot, now the company decided to perpetuity it, and it's going to be implemented in all eight campuses in India. The details of the mechanism, the experiment itself, the commuters participated, and the results show the people commuting in the off-peak time of by coming to work earlier than the rush hour, therefore being able to leave earlier than the rush hour increased markedly, the data will show that. Through the six-months about 1900 employees were rewarded. The average prize money per winner, the prize monies paid out varied in size, but the average is around $28, and if we had taken the pot of money available, paying as raffles, divided up evenly amongst all the people helping to decongest, each person would have made only 24 cents a week, would not have incented anybody to get up, half an hour earlier in the morning, take the earlier commuting time to work. However, the larger pay-out promised by a mechanism, felt Luke ra lucrative. This is the synopsis. The outer layer, called Bangaluru, most live in the oval section, to where the info sis labs are, and you see the funnel effect where the commuters come down to electronic city in the morning and leave to go back in the evening. Of the 14,000 commuters that participated in the experiment, 8,000 by bus, 3500 by cars or motorcycles and the remaining by public transport or live close enough to walk to work. As we look at the date available to us, bus data, in the spreadsheet every day, of the type you see at the bottom of the screen. Tells you, for example, the first line, bus number 54 picked up -- 18th and Main, arrived at the Infosys camp us -- 61 people in the bus, 12 people standing. In other buses there are spare seats, so for example bus 56 had four seats to spare. This is the data we got, commuting time and duration of journey, occupancy of the buses. High-quality data, fairly large bus for a large period of time. You analyze, in June of last year, some interesting and sort of, perhaps everyone disturbing trends revealed themselves. The top line, green colored line in the chart shows the total number of commuters that are part of the bus pool. This has grown from 8,000 to over 10,000 over the three and a half years. The red curve shows how many people were traveling during the later or congested times, commuters that were taking buses off at 7:15, essentially [indiscernible] lower traffic. The commuters are revealing a very strong preference for commuting in the later times. The raw number of commuters in the earlier times just went down. This was also true for all the other commuters, car, motorcycle, as well as the other IT companies. People preferred to come in later and later. As far as buses, the preference for coming in late led to redeployment of the fleet, more and more buses redeployed to come in at the later times. Gradually over time, the three and a half year period of the data, the number of buses coming in the late hours is around, I guess four and a half times more than the ones coming in early. So, if you now look inside the buses, the occupancy levels, you see this sort of red bars, how many buses, what occupancies, the two last tall red bars show you that over -- close to 70 buses have people standing in them, in June, the rides were fairly uncomfortable for a large number of commuters. This is the monthly average. The blue bars are about early buses, many running essentially empty. This chart -- various months, and here you see that the buses, from a popular origination point called -- the buses at uncongested time, 6:15:00 a.m. and 7:45, come in 30 minutes, the journey time is 30 or 40 minutes, whereas the 8:15 and -- pick-up buses took a very long time, average of 75 minutes, sometimes as high as two and a half hours. The longer journey times in the congested times, true for other buses, particularly on this chart. If we summarize the bus data we find that it's a long time and there aren't any seats for the commuters. We expect the commuters will change behavior or have some effect on commuter behavior, and it didn't, but we also showed benefits of commuting in the uncongested times. You can save commuting time by an hour each day, get more comfortable rides and there's a big savings in the fuel costs, per day. Now, for a whole month after revealing the data we didn't really observe shift in the commuter behavior, so we launched an incentive mechanism or program to pay out monetary rewards to help people change their commuting times to the uncongested period. The scheme itself looks like this. Depending on when the commuter arrived, if they arrived before 8:00 a.m. they earned a credit and a half and between 8:30 no credit. This is like having a frequent flyer account with your earned credits against your account, revealed on an employee website. Individual accounts existed, people could access, see how many credits they had. By the end of each week an incentive mechanism would take the credits of all employees and the money allocated for paying out incentive and make these random draws. Precisely how to random draws work, how much money was given out? This slide shows you how many rewards and what are the qualifying number of credits. Let's go over this bottom level here. The bottom level in the pyramid is showing there are 4800 -- sorry, 48, 500 repeat prizes. Either 10 or 12-dollars. In order to qualify for this bottom level in the pyramid a commuter had to have at least three credits in their account. If they had at least seven they qualified to the next level of the pyramid, at which level there are 12, each worth 12 12 rupees. If you consider what 12,000 rupees means, it's around 250 to $300. [indiscernible] the monthly salary of the employees in the company, certainly the bus commuters, and that's a fairly significant amount of money from their perspective. This is the pyramid, animation you will see of the draws. Our goal was to ensure that those commuters who had high credits stood to win more money, certainly true because of the way the qualification is set up. It is the case that if you had lots of credits, then you got to, stood to win a fairly large amount of money. What we wanted to do was more than that, wanted also to give a person who had high number of credits a better chance of winning. The way it works now is, let's assume this week three people qualify for the top-most level in the pyramid, only two prizes, so two out of the three are given the prizes and the remaining person is automatically qualified for the next level. This level happens to be more prizes than people who qualify, never happened in practice, this is a case for the algorithm. Here, people get the prizes and the left over money goes down, so it continues. You see how the more credits a commuter had, the higher the chance they won a prize at all. So, the scheme ended up rewarding systemically the high credit holding commuters, but don't disregards the commuters who decided after three months of observing the pilot, and being in effect, they decided they would start commuting at the uncongested time, they had something to shoot for. So this is the scheme. I am ignoring a lot of the details as to how the credits are maintained, how much is subtracted at the end of each week. Happy to tell you offline or point you to a paper on this. Let me tell you about the results. If you look at historical data, the dots on the left, from 2008, March, leading up to the pilots the week before, this number here, September, third week. There were 13 or 1400 people coming in before 8:00 a.m. The number launched before 8:00 a.m. increased, essentially up around to close to 2000, and continued that way until the pilot ended in April. After the pilot ended in April the numbers began to come down a little bit. That's also true for the number -- the red bar, and the people arriving before 9:00 a.m. This is how the response has been to the scheme. If you look at more aggregate numbers like person hours saved, for the bus commuters over the course of the experiment, this is what we see on a month by month basis. The average commute time, what's shown in the Y axis, came down from 70 and a half minutes or so, to about 54 minutes, the average commute time, and this averaged across everybody that's, everybody shifted times and those who haven't. After the scheme has been removed in April, April 10, when the scheme was stopped, we noticed that the recent data is showing that the commuters have gone back, there's a trending back to the late arrival times. Okay, so, all the data is one thing, but this for me is the most interesting and telling chart. This shows that the commuters that were demanding earlier, that buses be redeployed to later times, now after the scheme got lost demand buses be deployed at earlier times. Commuters complained about how slowly drivers were driving, asking us to replace them. We had complaints from the drivers saying what happened? They will normally come sleep, not care about the commute, now they are fairly vocal about getting their earlier. So, I guess [indiscernible] -- okay, so this chart is quite a strong indication, the commuters lobbied us to advanced bus times, to -- commuters complaining about falling in the cracks, not being able to get the credits. Several buses moved by 15 minutes, for the tall bar. The last bar, rescheduled, shows how many buses that were previously arriving after 8:00 30, so all the commuters being could participate in the scheme. These are the results I will show you, the most, leading indicators. Now, a summary of the presentation, the take-away from us from the project, argument in favor of a non--toll type of approach, one where you charge congesters, but reward non-congesters. More people willing to voluntarily change commuting times, perhaps. This is one example of it. Now, as I told you earlier, [indiscernible] is going to deploy the scheme in the offices. There's an interesting next step for us at Stanford University, with the parking and transportation services at Stanford University, going to launch an incentive mechanism, details in the next slide, but that's something that we are looking forward to, because it's in the same time zone, we can really participate in the experiment with [indiscernible], and of course there are other examples of entities interested in the instant experiment, the electronic city association, EL CIA, interested in trying us out, have the same common problem. The other is this Bangalore metropolitan transport, and other places of the Indian bus and trains being overflowing with commuters, people sitting on the roof, what have you, hanging off the [indiscernible] is actually true. This experiment, the public bus system wants to load balance the buses by trying to incent people to travel at uncongested times. Now, at Stanford there's a [indiscernible] for reducing -- this is the back of a campus shuttle. The image on the right is on a parking structure. These sort of signs are pretty commonly present on campus, fairly ubiquitous. The reason is that Santa Clara county, where Stanford sits, and Stanford, have an agreement that strictly limits the number of [indiscernible] during peak hour. The most congested one hour in morning and evening. There's a precise number more morning, 3150, the number of vehicles that can The campus, a 1% tolerance, and if the numbers are not met there are severe penalties. This is the amount of congestion that occurs on the roads leading into and out of campus is really what this is aiming to address. Now, the university has various measures that already address the problem, for example the university gives out free bus and train passes, gives out something called the clean-air cash, if you don't obtain a parking permit, you get money directly. So various things are done to encourage people not to drive or at least drive at the off-peak times. They are all fairly open loop, not measurement driven, it's not clear exactly how much of the effectiveness of the schemes is owed to the offers themselves. So it's not very are clear, there's not the direct relationship between the schemes and the effect other than that you observe something. What we are proposing, and we are going to work with the Stanford parking folks is measure the arrival and departure times. We will move from the standard windshield parking stickers, to RF ID park being stickers, and you can scan at entrance and exit points, in the parking garages, report commuter arrival and departure, and people accrue credit, especially in the off-peak times. The credit accrued can be refunded as part of parking fees, you acquire a fee for some amount of money, around $700 to get an A grade permit at Stanford, and perhaps two or 300 would be returned to you. Just an idea. This can be coupled with random draws where some people may get more money given back to them, and there are of course other benefits that a smart parking infrastructure -- [indiscernible] parking stickers create a smart parking infrastructure, and other benefits such as commuter parking histories that helps with capacity planning, differential pricing of parking lots, much easier to enforce parking. Parking, always people trying to park without permits, and easier to renew, so on. Over all, the parking folks are happy to try this out because it's really been a difficult thing for them to know how effective the various measures have been. We are looking forward to trying out the incentive approach here. That concludes my talk, thank you for listening. I will turn it over to Darryl. Thank you very much. Next we will hear from Nicholas Ramfos from the metropolitan Washington Council of Governments. Thank you, today I will be talking about a few incentive programs being used in the Washington region to help reduce congestion and air quality. I want to give a brief background of commuter connections, it's a regional wide program, operated in the Washington D.C. area, local government, ride share, federal agencies, transportation management associations that work together to provide demand management program services to the general public and business community. Commuter connections, TPB is MPO for the Washington D.C. metropolitan region, housed at the Council of Governments. In other words to meet air quality goals, the TPB adopted emission reduction measures, terms used to off set and constrained long range plans. Commuter Connections adopts guaranteed ride home, telereach and mass marketing. Each of these terms have set emission and transportation goals. The first regional incentive program I will discuss is the guaranteed ride home or GRH. We launched one of the first guaranteed ride home programs to the idea to retain commuters in alternative modes. The program positioned as an insurance program for commuters who experience an unexpected family emergency or unscheduled overtime. The very rigorous evaluation is used to determine the effectiveness of all Commuter Connection programs an analysis determined benefits for the region -- data collection results show the guaranteed ride home attracts users, and more than half the participants had been using an alternative program less than two years. The program attracts long-term users of alternative modes. You see impacts on the transportation side and emission side. I won't go through all of these, but we account for vehicle trips reduced, vehicle miles of travel, NO x and [indiscernible] One telework partnership with the [indiscernible] in Maryland, and the telework! VA program. Offering consulting services to employers seeking to start or expand a Telework program. The programs have offered hand-holding to employers unfamiliar with telework and provide employer with tools, resources for implementation of or expansion of a telework program. To date several employers have taken advantage of these incentive programs. There are associated emission impacts for each program shown here, both VOC and NO x. The final incentive program I will discuss today is in the process of being developed, and it's a carpool incentive demonstration project. Last year Cox commuter -- several areas already operating pool incentive programs and the feasibility study took a close look at each of these programs, in Atlanta, New Jersey, to determine which if any can be applied to the Washington region. We will begin this fall, and the cash for commuters program, along with the LA ride share rewards program. The benefits from the Atlanta program include a 24% retention rate, and 64% after 9 to 12 months of participation. In the Los Angeles evaluation, the ride share rewards program, 52% remained in a carpool after the program ended. For the Washington D.C. region carpool incentive program, the top 10 congested corridors were selected for close ex-am nation to determine what type of road ways or corridors would most benefit from the incentive program. Criteria for select thing the corridors were identified and included in the feasibility study. Three corridors were ultimately selected for the demonstration project, and incentives in order to -- from sing occupant -- they don't have any HOV access, either Program participants will receive $2 a day for their participation, over a 90-daytime period, their participation will be self-reported, electronically electronically, eligibility by the online application process. Both of these methods are consistent with other carpool incentive programs reviewed and currently in existence. A survey will be conducted at the end of the program participation period in order to gouge effectiveness, and the results will be used to determine if the program will be used or expanded in the region. That's essentially my presentation. Thank you. Thank you very much. We have a whole bunch of questions queued up, so I will paraphrase and run down them. Firsts, for Dr. Prabhakar, where did the money for the lottery funds come from? Was there a congestion pricing fee assessed on peak travelers to fund that? Dr. Prabhakar: No, there was a pot of money available in the -- the commuters obtain monthly bus passes by purchasing them. That bus pass fee pool had a surplus. That was made available for the pilot. Looking forward, we are going to -- the company is going to pay some money and the remaining money will be from the commuters registering into the program. It's a combination of commuters paying themselves, and everybody pays up front, only the early-comers, or people traveling in uncongested times gets the rewards. There's an interesting aspect, the amount of fuel saved by moving the buses to off-peak time. One thing I didn't say here, the better load balancing of the buses led the company to remove 10 buses from the fleet. That's a permanent savings in money, that can be applied to the scheme, people are sort of figuring these things out, while -- from the running of the scheme itself. Next question. Here in the U.S. we are very sensitive to travel times. There's a high degree of sensitivity in patterns to the time savings. Why was -- or can you determine was that time of savings not enough to get people to switch their patterns in India?