issue of work hour limits, and we still have in the united states people working, as we heard related to the police officers, working extended duration shifts. we know in the medical area they're working 24-hour shifts. we know from our work with firefighters that they're working 48, 72-hour, even 96-hour shifts in some cities in california. and many times they're able to get some sleep during those shifts, but on many occasions they can't because they're dealing with paramedic responsibilities and others, and so there's no real provision. we heard issues about could transportation be provided but most employers don't provide those transportations. whereas the person may be driving a commercial vehicle like a fire truck during their work hours, they're going to be driving their own vehicle, a noncommercial vehicle, driving

home from those extended duration shifts. it's been recognized in the transportation industry since 1907 when in the united states congress passed legislation saying anyone working for more than 12 hours would be considered -- that it's illegal and the supreme court has ruled that they're considered impaired by fatigue if they're operating a train for more than that time, and yet we don't have any limits, even though we now have limits for commercial drivers, train operators, bus drivers, pilots, mariners and so on, there are no limits on work hours in so-called nonsafety sensitive industries, and yet in those industries, as we heard today, in those industries people are driving home many times for doing super commuting. we know from our own research that there's 168% increase motor vehicle crashes after working extended duration shifts. we have no limits in the united

states for those work hours, and i think an important countermeasure would be to consider limitations such as the european working time directive that says in all occupations work is limited to no more than 13 consecutive hours, and that covers all 330 million people in the european union. another provision of that is that there has to be 11 hours time off between shifts. again, covering the entire population. and that effectively limits work hours to no more than 90 hours per week where as in the united states in some of our studies we have seen people working more than 140 hours a week. the concept of compensatory sleep and through provision of controlled rest in the workplace needs to be considered as a countermeasure in many other occupations. anthropomorphic measures is

another measure that can be used to identify drivers and require objective screening not just for commercial drivers but also for those who are getting a regular driver's license. and it was brought up the concept of requiring physicians to report people with sleep disorders. a very exciting countermeasure we heard today was center line and shoulder rumble strips which -- and roadside rest areas which sound like they could make major improvements as well as in vehicle systems to identify those who are drowsy -- driving drowsy and to initiate change as the driver is operating the vehicle.

i want to end with mentioning drowsy driving legislation. the national sleep foundation supports the drowsy driving act of 2015. the foundation also supports the "z"s to "a"s act to provide guidance to schools regarding school start times and the american academy of pediatrics as was mentioned today has recommended that no middle school or high school in the united states should start before 8:30 a.m. and a number of studies showed that would decrease drowsy driving crashes among teenagers. in massachusetts in terms of models, we have required drowsy driving education for all junior operators as part of a junior operator law passed in 2007 and not only prohibit but have greatly increased penalties for

junior operators driving at night. as a result that has resulted in a 19% reduction in crashes. 40% reduction in fatal crashes, and a 30% reduction in nighttime crashes. so thank you very much. >> thank you. thank you for the comprehensive review and the countermeasures that you put forth as particularly effective. our next speaker is mr. jake nelson from aaa. >> thank you, dr. price. so i have a lot to get through, and i kind of approach this a little bit differently in that there are several things that really stood out to me from the day, and i'm kind of a guy that likes to think so what do we do about it? and so the theme for the day for me was what can we do in an environment where there is no substitute for sleep, yet the average motorist is very cavalier about this problem?

so my recommendations fall into three categories, data, prevention through education, and then safety net interventions. and before i start, i just want to say that i'm making some of these recommendations in some cases without knowledge of what might be going on in some agencies already. so first in data, it was said right off the bat that state reporting on fatigue is weak and inconsistent. i would say it's fairly obvious that if that is the case, which we know it is, that's going to roll up to the national data as well. so to the help address that problem, i think nhtsa and isap could work together to help better identify fatigue as a role or factor in crashes. we had an example of crashes without skid marks if for no other reason to suggest it might be worth looking at fatigue as a potential factor and clearly there must be other items to look at there.

outside of the police reported crash data, it seems to me that nhtsa has an opportunity if they haven't already through its data modernization process to enhance the size and scope of cds such that imputation could be used to analyze those data the way it is for alcohol, to look at the issue of drowsy driving and its role in traffic crashes. and then also i think just -- and this applies for traffic prevention generally and drowsy driving would be no exception, nhtsa ought to initiate development of a system for injury crashes across a range of severity. i know there's an nchrp project already under way looking at developing a framework for doing this, and i think once that project is done, i'd like to see it put into action and not just sit on a shelf. there is a road map that is coming our way to help address that issue.

and the world of prevention through education, for me the first panel really underscored a point which was that drowsy driving may not even know that they have been asleep at the wheel. which tells me that primary prevention is really important when we're talking about addressing fatigue driving. i think that given the implications of fatigue on injury, both traffic and other sorts of injuries, workplace injuries for example, and then also health issues like obesity, the cdc ought to be developing interventions for state and local health departments to educate consumers at the community level about this issue and that to the extent it's possible and makes sense to do so work with nhtsa to help get that information out, the same interventions, the evidence-based interventions out through highway safety offices in the state. a focus on high risk groups for those interventions makes sense, teens, shift workers, elderly people not only diagnosed with sleep apnea but at risk of being diagnosed with sleep apnea, and

also one more point there, that i think it remains to be determined if the shock and awe way about educating people on this issue is the way to go. i'll leave that to the experts at the cdc to figure out. also in the area of education, i think that, and we've talked about it a lot today. i think it's necessary but insufficient to address the problem. i think it's absolutely critical that we provide education in a variety of contexts, but i don't think that education by itself will fix the problem. i think people need options to tap into if and when they find themselves in a fatigue driving situation. if we're lucky through education to have gotten them to a point where they can recognize that's something they're struggling with, if we don't give them something to act on, i don't think we've solved anything other than they know they're in a dangerous situation but they don't know really to what to do about it. so some examples we heard today were safe pull off locations or better signage to direct people

to where those can be found. employer programs like controlled rest programs, point of decision prompts when it comes to drugs and how that can affect a person's ability to drive safely, nonpunitive environments to get or access treatment without fear of job loss when it comes to things like sleep apnea and other sleep disorders, and i would also bring up the whole issue of referral by law enforcement and health care providers to state licensing agencies when it comes to medically at risk driving and drowsy or fatigued driving would be no different in the context of a sleep disorder. i think there's a lot we could do there as well. bottom line, we need to ensure that motorists are enabled to leverage these resources, not just to be aware of the fact that drowsy or fatigued driving is important and that it's something they need to be aware of. in the area of safety net interventions, i think that vehicle technology should and probably will without us doing

anything aid motorists in detection of fatigue. there was some discussion about advanced driver assistance systems. i think it's pretty well accepted what the potential for these technologies is to prevent crashes, fatigue crashes, distracted crashes, all types of crashes, and research that the aaa foundation has done working with the mit lab, while there's a lot known about the potential, there's little known about the real world effects they can have. we know about efficacy, very little about effectiveness. so we need to focus on that. and then the infrastructure related countermeasures can be low cost and very effective. we've struggled with the issue, the rest stop issue at aaa over the last several years because we've seen state financial pressures result in those being closed or the pursuit of those being closed. it would be great if there was some sort of resource to help state policymakers sort of understand everything that needs to be considered when tackling or trying to address that issue

so they aren't identifying it and trying to legislate on it in a vacuum without considering all of the other implications that might not be in the front of their mind. i think i will stop there with 14 seconds left. >> thank you very much, mr. nelson. our next speaker will be allen pack of the university of pennsylvania. dr. pack? >> okay. well, thank you very much for inviting me to this. this has been an extremely interesting, informative event. let me just give you my big picture conclusions. the first thing is i thought the talk dr. taft gave on the scale of the problem was probably one of the most balanced talks i have ever had on the different approaches, and his estimate,

which he directed to you, dr. rosekind, of 5,000 deaths and 100,000 injured sets the scale of it. i think the other aspect of this is that we've -- it's a hard problem to get around because it's a very diffuse problem. it's very diverse, different groups here and different groups there, and one solution doesn't fit all. it's a very diverse problem. it's not like you're dealing with commercial aviation which is a much more focused thing. so i think that's the second point i would want to make. i think, you know, you can look at this and say have we done anything over the last ten years? i think there's a lot to be done, don't get me wrong. i think there's a major amount to be done, but we have made some incremental change. i think the slides dr. watson showed on the guide to a driver's license and what that has done, i think the relentless effort to change the school start times, these are all examples of progress we've made.

so we are making some progress but there's a lot to be done. i think this is actually an extremely optimistic time because what you've got now is you've got a number of -- the ntsb obviously is very important in this regard, but you've got nhtsa, you have got fda involved, cdc involved. then you've got a number of non -- national sleep foundation, american academy of sleep medicine, cdc. there's a lot of players in this game and aaa. there's a lot of people in this game now who can coordinate their efforts and i think we can really make a difference. so i think coordination is a very important part of this moving forward. now, to return to the scope of it, i think getting a handle on the scope of this is much more accurate, is actually quite important. we've heard numbers from 829 deaths to 5,000. 829 is unacceptable as well, but there's a big difference between 5,000 and 829, and i think we

all believed and have believed for a number of years that the data and the federal data we have captured really massively underestimates the scale of this problem, and i think dr. taft's talk indicated that. and that's not just for me an academic enterprise. i think one of the things that alcohol has had is very good measures. if you're investing in education in alcohol and you can document that you're seeing, you know, crashes due to alcohol declining, then you're getting encouragement, you're getting a return for investment. and so we talked -- we're talking about education, but let's say you put that program in place in utah. how would you know that that made a huge difference if you're relying on unreliable data? so it seems to me having reliable data on this is critically important. and i think that's a role that the ntsb can really help with. dr. taft had an argument -- not an argument, had a proposal about in-depth analysis.

it seemed to me that we have technology right now, at least very close to having technology, that if it was in every car, if there was a crash, i could pull over and i could say was that driver driving drowsy? i'm not talking about giving warnings to the driver. it seems to me there's technology you could put in cars right now or pretty close to it that could actually tell you at the scene and instead of the policeman ticking a box when they're poorly trained, you would have data to look at and it seems to me that's quite important to consider. so getting a handle on how do we better estimate this, better know what's happening i think would be very important because it would encourage the educational investment because you could document return on investment. and the other thing, the other concept i got out of this was there was a tremendous variation between different segments and it really was quite impressive in that panel where you've got bp saying we don't care about costs.

this is our volume. we're going to do this because this is our value system and then you hear about the police who don't seem to care about this even though they're the ones out at the scene seeing the events. just inconceivable to me. and understanding the culture and understanding the barriers in these different segments seems to me to be a very important part of research. and in terms of the clinical, i think what the clinical panel clearly showed is there's so many dimensions to this, short sleep, being up at night, this disorder, taking drugs. there's so many dimensions to this problem and getting your hands around it is pretty challenging. i think as was said the data is most compelling for sleep apnea. that's because that's where the studies have been done. dr. rosekind asked the question, we really don't have that type of data. there's multiple studies and you're focused on that because that's where we get the data. over here there may be more

problems we just don't know about and i think that's tremendously important. i think the idea that all of the pharmacists playing an important role here is a great idea. the other part that was talked about a bit was the alcohol sleep and i think that could be something that's expanded because there is data out there that if you have low level alcohol, you're not that impaired in driving simulators, slow level sleep deprivation, put the two together they massively interact. as the bac limit comes down in terms of risk, then the relative role of sleepiness increases. so i think that's important to consider. that interaction is actually i think very important. finally, i'd just like to make the point there seemed in all these discussions here to be a lot of emphasis appropriately on future research directions, but the research tends to be siloed in different pockets, and one of the things that i'd just like to mention briefly is that there is

something called the national center for sleep disorders research. when senator hatfield who was a person who appropriated the national center language, he envisioned the national center for sleep disorders research to be a coordinating mechanism not just within nih but across all federal agencies, so we have a coordinating mechanism. it seems to me we should think about using that and how do we develop a coordinated research plan moving forward? two of the areas, i'll quit in two seconds. two areas they're interested in is looking at biomarkers and can we get a measure essentially of sleep drive and how sleepy people are and the other area of real interest in biomedical research is what's called implementation research. i'm no expert in it but the idea is implementing obvious guidelines. for example, when the asthma group came out and said people should be prescribed inhaled steroids, the prescriptions for inhaled steroids declined after that. that was not a good thing.

and so there are actually barriers and it seems to me the concepts of implementation -- you're doing research, but then how does -- as dr. watson said, research without action is just academic. what are the barriers to doing that and i think a field of implementation research which is now happening in the medical world, i think what are the barriers we can change is an important area. so thank you very much. >> thank you, dr. pack. i'm so interested i keep forgetting it's my turn to talk. our final speaker will be steven popkin. dr. popkin. >> thank you, dr. price. so following on from the three other panelists here, i view that there's three issues, and this complements what has just

been said. one, there's a real culture issue if we're talking about sleepiness and people's willingness to forgo sleep. second, there's a lack of data, especially evaluative data, return on investment data for people to understand what the benefits are for different types of countermeasures, and, three, we're missing some key partners in this forum and in this discussion in general. before i get to that, dr. price asked me a quick question when she was inviting me to this panel saying what's the rest of the world doing? so i made a couple phone calls, less than nine, and just to give you a flavor of what's going on around the world. in australia, we're not talking about hours of service regulations here. we're just talking about are there rules for people driving their cars and getting into accidents related to drowsiness and fatigue. in australia, no, there is not. they have very significant road campaigns which could be used in court saying you should know

what's going on with fatigue and you should have been aware, but they don't have any particular rules about how many hours you can drive, for example. they do have what's called a duty of care, so the employer needs to educate its workforce about driving while drowsy, but they don't have anything beyond common law and criminal law specifically on drowsy driving. in brazil they don't have any laws for people driving cars or motorcycles. if you're driving heavier types vehicles, there are some rules that you have to be checked for sleep illness, but if you're just driving your own car or a motorcycle, there are no rules. in canada similar to australia, they have a lot of signage on the roads, drowsy drivers, there's an exit in the next five kilometers down the road but they sort of lump this into

distracted driving, that a driver's judgment is compromised when not fully focused on the road. that's how they look at it. in germany there is a rule that sort of lumped together you can't drive while impaired and they look at drowsy driver as impairment. if you get caught it's a 25 euro fine. if they figured you're endangering life it's 10,000 euros but there's no regular controls. there's no way for the police to really understand whether you're impaired or not, so very often i was told that the police officer would say go home and get some rest and there's no fine whatsoever. in the eu, which is a super set of what goes on in germany, after nine hours you're supposed to stop driving. thereafter, you have a 30 or 60 euro fee per every half hour thereafter. and one more, in finland you should not drive while you're

ill or tired or otherwise impaired, and that in some ways falls into the dui law, too, where if you're in grave endangerment of the traffic, but in general there's no good data from any of these countries or understanding about how well these laws or the communications campaigns they have for the signage on the side of the road, any of these techniques work and that these laws are often lumped together with other laws, outcomes, and they're not well enforced. i think we can take it that the united states is not lagging behind where the rest of the world is here. this also makes me understand that sort of laws and technology are no real panacea for behavioral change. we have a low signal to noise ratio for the behavioralist in the room. you can drive drowsy a lot of times without anything really happening. you can pass police officers a

lot of times while you're drowsy and you're not getting pulled over. if you're pulled over, nothing really happens to you. so there is no behavioral change based on the system that is currently in place. i also think there's a low acceptance and implementation of laws. it's hard without any sort of -x5z physical evidence with cell phones, you can see somebody is on the phone or that they're talking and nobody else is in the car. you don't have that, and so it makes it very difficult for the police to be able to identify you if you're not wearing a seat belt, they can see that. if you're tired, it's much harder for them especially if you're just driving by. and lastly, i'm concerned about inappropriate reliance on technology and communication. even if these in-vehicle systems do work, are we going to become overreliant on this?

what are the unintended consequences? for telling people about fatigue, are we then brushing our hands saying, okay, you know about it. this is an old trick that used to happen in the shift work industry. here is a manual about fatigue, do you understand it now and it's now shifted the liability from the company or entity to the individual. i have three takeaways, still a little bit of time. one is data, and we need data, one, on understanding the problem but also understanding the safety impacts, the return on investment. if we're developing these countermeasures, are they really making any impact? we heard about, you know, getting leaflets in the medicine you take. we don't know whether people are

reading that, whether they understand that well enough to really act and make decisions on that. that's key. we know from helmet law that is you could have all the data in the world showing the value of wearing helmets, but if people don't buy into the message when they get to the massachusetts/new hampshire border, they will take their helmet off and keep on moving. you have to understand what's important to the people. you got to make it about them and maybe that's through making it a public health campaign, but you have to get them to make it about them and their investment. second, you need to build a community with a notable champion or champions, and i think if we're talking about the entire population, we're going to need multiple champions for this, but people who you can relate to. former secretary lahood was distraction. we have michelle obama who is for obesity. michael j. fox for parkinson's, but you have to have a person, a human being. people don't relate the same way to an organization saying you should be getting more sleep.

they may provide the credibility and foundation for that, but you need a person. you need somebody up there that people trust and respect to relay that and to build that community and to relay those messages. and lastly about communication, i think we can learn a little bit from daniel, thinking fast and slow. we have the dichotomy. you're thinking slow when you're thinking about what are all the benefits for getting a good night's sleep. what are the consequences if i'm not going to get enough sleep? but then you have the thinking fast, the emotional impulse of, you know, i love to have the wind in my hair. i don't want to wear my helmet or i want to spend time with my friends or, you know, i really want to get a good grade so i want to stay up late at night. so i think if we're trying to develop communications, this gets at a point of who is not in the room here who i think would be good, behavioral economists or people from madison avenue. people who know how to segment a population and communicate to that population and get them to change their behavior. at the end of the day i think that's what we need to get to. we can have all the rules, all

the fatigue risk management plans, all the technologies, but if people aren't buying into this, they'll circumvent it or won't pay the type of attention they need to really make the type of change we're talking about. 5,000 deaths is unacceptable. one death, you know, something that is so easily remedied is unacceptable. we need to find ways of communicating that and while we're great at the science here, we need to bring those people who really know how to communicate into the conversation. thank you. >> thank you, dr. popkin and thank you all of our presenters for your phenomenal presentations. really amazing food for thought for us. i would like to ask a few questions and i would like to start with a theme many of you touched on and that's data. we've emphasized the importance of data not only for looking at the prevalence of this problem but also and perhaps more importantly looking at the effectiveness of countermeasures

to know whether the things that we think are solutions are actually working. i have firsthand experience as a human performance investor in major transportation crashes in looking at fatigue in crashes in that it is not easy, it is not easy at all even for a trained investigator who is given a free rein to look at everything under the sun, even for our investigators it's challenging. so i have to ask, i think we heard some very interesting suggestions on way that is we can improve the data. i heard mr. nelson say things about training law enforcement better, enhancing the size and scope of the data, using imputation, expanding f.a.r.s. we heard recommendations, dr. pack mentioned could cars tell

us whether people are driving drowsy at the time of a crash perhaps, and we even heard about other things like biomarkers perhaps in the future there will be a so-called fatigue-alyzer that will tell us whether somebody is involved in a crash. i'd like to open it up to all of our presenters to ask do you have any concrete suggestions for ways we can look to improve this data so we can help to really do a better job of tracking countermeasures and understanding prevalence? and i open the floor. >> i would just say that even just the few suggestions that this panel has made for at least to me seem to sort of fall into a time line. i think working with what we have is a near-term solution. so the ways that mr. taft at the aaa foundation has decided to study fatigue as it relates to traffic crashes is making the best of what we have, and that's something we can do straightaway.

the discussions about biomarkers and having technology in cars that can tell us at the time of the crash whether or not somebody was a fatigued driver are also really great and out of the box ways to get at this problem but they're probably more in the future. so i think they're all worth pursuing and they probably will have more value and utility at different points in time. i don't think that we should wait to do anything to wait for things like biomarkers and technology in cars when this is a problem today and we need to do what we can to understand and measure it as best we can today with the information that we have. >> thank you. does anyone else want to chime in on that, dr. popkin. >> i will chime in in two ways. one, if you're developing a comprehensive countermeasure or approach to trying to address

the problems, there is one opportunity for you to build in an evaluation approach to understand who are the stakeholders, understand what they care about, develop the metrics, and build that into the process and the product that you're trying to put out there to make a change. this is something that in working with harvard we developed such a resource for in this case locomotive engineers and their spouses, but in doing that we worked very closely with a tight-knit community that they are to understand all the hurdles, everything that they would be concerned about and then built that into a continuous improvement process. so developing the data so when you're going through one of those, when you're trying to make even a strategic change in a particular segment of the population, you have that opportunity of really collecting some really good data. if you want to -- but if you're

talking about the population in general, i think the first step you really need to do is legitimize the topic among the population. getting people to have the nomenclature, getting people to understand the issues, otherwise it's going to remain underground, and i think we heard, you know, today building it up in the medical community, building it up in the schools, building it up, you know, churches where people come and talk about their problems, you know, and have a place to come commune and feel comfortable. i think you've got to bridge out from the traditional thinking of, you know, we're going to have research projects and we're going to collect data from that. i think you really have to get the population to understand what the issues and talk amongst themselves before you're able to get a really solid handle of that outside of being able to take a genetic test or something.

>> we heard today about many of the in-vehicle technologies that are being developed and that have been developed and that are being deployed in a number of vehicles that are already available for purchase, and many of those systems collect data and record them, and the manufacturers are collecting that data in some part to defend themselves in cases of lawsuits because often times when there's a catastrophic accident, the manufacturer is sued, and so they then have this data saying the individual didn't make any steering movements or whatever as opposed to something else happening in the situation that the brakes didn't work. well, if you didn't put your foot on the brakes, they're not going to work, so they're recording that information, and yet that information, although it's recorded and is it available in these crashes, it is not being -- we're not accessing that in terms of, for example, the f.a.r.s. could be

greatly improved if in every one of those fatal crashes the instrumentation already in the vehicle would be downloaded. so we have the information but it's not being systematically collected, and i think that's a tragedy because it's just going to make this problem continue unabated. >> thank you very much. and i want to continue with the countermeasures discussion. i was very glad to hear that all of you talked about what you think are some of the most promising countermeasures moving forward in some of your opening remarks. and i'd like to hear now about what you think are some of the most pressing near-term solutions you heard about today that you think are the things that we can look at doing right now that we're not doing. >> i'll jump in. i mean aaa has 44,000 employees

and over 55 million members in north america, and we have a direct route to them in terms of provision of information, and while i'm the first to admit that education is necessary but insufficient, i think that's a first step, and that's something we can do straightaway at aaa, and sitting here listening to the panelists speak today just occurred to me and reinforced over and over again that even as someone who is responsible for the safety programming within the aaa group within north america, i learned so much today about fatigue and driving and there's a lot i think that we can do to educate ourselves internally through our employees but also our members just on the topic in general as a first near-term step. so it's something that we're planning to do. >> i would say there's a couple things. number one, you know, the police

it seems to me is a very important group to focus on because they're the group out at the scene, you know, collecting the data and then obviously a pretty at-risk group themselves. having specifically problems for police and there was a paper published showing the problem of sleep apnea in police. having the focus on police as one important segment is really important. the second thing is clearly the sleep apnea issues we've had from dr. g, the prevalence of that is increasing as the obesity rates increase. it's not a hard condition to identity and it's extremely treatable and so i think that's a very important thing. given where we're at now in health care with emrs and so on and people filling out questionnaires and electronic websites, identification of this is pretty straightforward, and making that systematic, and i know the american academy of sleep medicine now has a task force talking to the emr

companies about can we get that stuff programmed into emrs. it's something that's extremely doable. >> would anyone else like to chime in? >> i guess i would just proffer a little bit of a warning to whatever goes on needs to be very judicial. you're talking about what's going on in people's bedrooms literally and in a lot of cases their livelihoods. so when -- and this is the same as when you're going in looking at a shift work organization. when you start telling people that fatigue -- we're going to start cracking down on laws or your car is going to snitch on you, whatever, you might get a really big backlash from people saying i don't want big government going on in tracking me. i think a soft approach, even though it's to a difficult

problem, at least to start with is what's needed. through education, the aaa foundation. aaa reaches 55 million people that's a whole lot of people to start educating and to start giving them the understanding and, again, you know, doing a little bit of research understanding what are people going to resonate to? are they going to -- i would think maybe they'll resonate to i want my son or daughter to do really well in school or to be really good athletes and now i know if they don't get enough sleep, that's going to be a problem. so understanding where people's decision points are rather than being somewhat heavy handed on a regulatory front is my warning. >> as you've indicated, to do this properly, you need real professionals. you talked about madison avenue types and really developing proper communication strategies,

and so the investment you need -- i mean, it was mentioned the magnitude of the investment in alcohol education. the investment you need in this to make education work i think is extremely large and is going to involve real professionals doing it, and the question then is how do you get that level of investment? i'm not opposed to doing education but how do you get the level of investment you need to make a difference and how would you show that that difference was actually occurring? >> mr. nelson or -- >> i was going to say one thing, when you talk about the investment and the madison avenue folks getting involved, you have to think about what dr. watson pointed out with this caffeine industrial complex as he called it. and that is that the madison avenue folks, production and consumption in our own economy is fueled by sleep loss in many senses, and so the madison avenue folks are actually trying to make sure that, for example,

currently there are 3 billion hours spent a week video gaming. by 2020 the video game community wants 20 billion hours set on gaming. their sights are set in the reverse direction. in order to be effective at changing this cultural notion that sleep is optional and it's trivialized and so on, we have to recognize that it's not just a matter of providing the education but recognizing that there are, as dr. watson pointed out, forces that are providing the reverse information and in order to be effective, we have to be able to counter some of that information. now, with respect to the car snitching on you that dr. popkin suggested, what i'm talking about here -- that would be like the airplane black box is snitching on you when it's looked at after a crash. i'm not suggesting that it be

looked at every time they fly and i don't think the national transportation safety board looks at the black boxes after every flight but when something happens in those 30,000 or more death that is occur on the highways annually, we could look at the black boxes when something catastrophic happens, when there's a debilitating injury in order to learn what fraction of these are being -- are likely to have been caused by fatigue. once somebody is killed, then i don't think we would view that as the car is snitching on you. it's taking information. right now most of the time nobody looks at the black boxes in those situations even though the information is being gathered and that i think is the problem. >> just to respond to that, i think there's a difference between flight pilots and people driving their personal vehicle, one is a very controlled industry and one is very deregulated.

but my point was predominantly that it's a culture shift. people i think now would not like the idea that somebody is going to pull data off and it's going to change my insurance rates or what have you or i could become more culpable. i think where we want to get is to that can happen and we can use those data for understanding what happened in a particular crash, but at this point i think there may be a backlash. i could be wrong. >> what's the philosophical difference between somebody at the scene giving you a breathalyzer or somebody at the scene taking -- looking at what's in your car? what's the philosophical difference? the driver is there, i say you have to blow into this or i'm going to take the piece out of your car. so what's the philosophical difference? >> i think we've moved along from short-term solutions to actually the group has intuited my next question and my final question, so i will ask it and

i'll allow you to continue a little bit of the discussion you're having and then we can turn it over to dr. malloy and member rosekind. as an aside, i will mention in aviation, the airplane black boxes sometimes are not by ntsb but sometimes are by the industry used in an aggregate form to look at safety issued by deidentifying it from the individual. so there is exciting opportunities potentially there, but certainly aware of the warnings that dr. popkin is making with respect to privacy concerns. the final question i wanted to ask has to do with an issue that dr. popkin did raise, which is implementation barriers and how can we get past some of those barriers? i think we're hearing about one right now, the concerns about individual behaviors, individual privacy concerns, and i'm sure there are others that you'll want to mention.

are there other barrier that is you see as being particularly concerning for implementing some of these measures and if not, that's fine, we'll move on. >> i think it really depends on what group you're looking at to understand what the barriers are there. for example, when we were trying to address drowsiness in the railroad community and we wanted them to have access to a sleep disorder screening tool, the railroad management was all for that. the railroad laborer said no way unless there's going to be absolute protection that whoever uses this tool cannot be traced back and the results of them using that tool cannot be traced back. so that's just a small section of about a three-year

negotiation between the regulator, the laborer, and the management to get that project going. so i think when you actually are going in and trying to understand the population you're changing, that's when you have that stakeholder engagement. that's when you will learn about a whole set of issues you never really thought about before that could be hurdles or derail the behaviors you're trying to change. so there may be some that are more global, but i think the devil is really in the details here when you really try to understand who you're trying to change and what their view of the world is. >> i mean, it seems to me that there are perhaps general implementation barriers and then there are very specific implementation barriers. we have the culture of the police for example. you can think of it as a general thing and then specific, and what i believe is that we need to build here the same type of

research that's going on in the biomedical research community which is this implementation science or patient centered outcomes research and we need to think about that because that's a whole branch of science. i'm no expert in it, i just know the terms. when the patient centered research institute which was funded under the affordable care act, when you apply for grants to that group, you have to get all the stakeholders together. you have to have patient groups, you have to have insurance companies, you have to have health systems and patient advocacy groups and so on, and then you decide together what the question is. so i think there's a lot we can learn from that type of activity, the patient centered outcomes research. they're heavily focused on the idea that we've learned a lot from academic studies but if they're not being transferred into improving people's lives, what's the point?

so i think we can learn from that, i think we can learn from the strategies that they're employing. i think we can learn the strategies that they're employing, import them over here and really try to build that type of structure to understand the barriers and how to reduce them. so i think the strategies we could adopt that we can employ in this area. >> thank you. i want to make sure there's time for other people to ask questions, so i think at this time i'll turn it over to dr. malloy. >> i just have one question for the group. we have been trying to change behavior in a number of ways with drivers. i'm thinking of seat belts, impairment, distracted driving, and one thing the board has consistently gone to is we need education, we need strong laws, and we need enforcement of those

laws to get behavior to change. how does that model not necessarily work in this situation or does it work? >> i think that that model does work. when we were working with state senator richard moore in massachusetts on the drowsy driving legislation which led to the junior operating law, he said we need to have education and drowsy driving legislation that he introduced there provides for a two-year implementation period where education is provided. but then he said if there is no legislation behind that that says if you've been driving for 24 hours and you're involved in a wreck, this is similar to the statute under which the driver in the crash that injured tracy morgan has been charged. but, unfortunately, new jersey is the only state in the union

that has such legislation. that's why national sleep foundation is -- has sent out information to all 50 states to get the driving act of 2015 passed. if you don't start with education but have some enforcement center -- if there's no consequence, then people are not going to pay attention to the legislation can you imagine? if we had no laws against driving under the influence of alcohol, that $300 million that's spent would be much less effective in trying to reduce the occurrence of drenging and driving. but i think that there are also some things that can be done here that are -- as dr. papkin would say softer ways of approaching this problem that could also be affected. so for example some insurance

companies for both and she shall fleets as well as individual drivers are offering insurance premium reduction incentives for driving the vehicle that can monitor whether or not individuals are falling asleep at the wheel. and, so for example, here in washington air e area, the wamada, the washington metropolitan transit authority is continuously dealing with the fact that drive cams are showing that drivers are falling asleep at the wheel of their various vehicles. and so that data as dr. pak would say, is sort of immediately has to be kelt with. and, similarly, if insurance companies offered those incentives widely and more people install those types of technology in their vehicles, then it would be easy for both

the individuals themselves to recognize that there's a problem, as well as for the met data to be collected and provide the kind of information that's needed. there is no reason that that approach that you suggested should be used here. but i think other things can be done, as well. >> i think the challenge to that approach, and i understand the model, but i think the challenge, it's just what we're talking about earlier. that is the difficulty in measuring the roles of drivers in the crash. you have a bac, the physical number and the curfews don't always work, you can see a young person driving around at 2:00 a.m., they're pretty ovts ebvious. i think the challenge in the regulation and the legal is can you measure the outcome. and it's enforceable. and the out come it just seems is going to be quite challenging. and you could end up with loads

of lawsuits and people making loads of money and not making a big difference. so i think the ability to measure this thing is absolutely critical. >> i would agree with that. i think it's probably the reason why only new jersey and arkansas have policies in place that address this issue. i don't think it's easy for people to grapple with. i think the enforceability issues are insignificant. it's not a reason not to pursue policy, i'm just not sure that the public is there yet. and maybe education is one of the ways that we can get the public there. if we can get the public to understand that this is a very risky thin to do. maybe folks would become much more open to policy folks starting to address it. >> i think that -- gist tojust to respond to that -- i think the legislative approach 30 years

acould would have beena ago would have been difficult. but, today with all the technology, it's very easy to immediately construct a sleep history just as the national transportation safety board does in incidents related to -- and people are wearing these monitors and so on. for fit bits and job ones and so on and so on. and so the data are increasingly available that could be, again were the municipality. so if the law was on the books and then you go gather the data in the case of fatality, we're not talking a huge number instead, you have to have -- although i'm very much in favor of research to develop the market to this.

>> that would capture one dimension of this. that would capture the person with acute sleep problems. >> we'd be one step much closer. >> thank you very much. clearly, that's a debatable option there. >> superb handle. >> given the time, you get one line. i'm curious, after today, if you're writing tomorrow's headline, what's the number one take home message you want people to know about this issue, this forum. so that's one line. not a closing statement. and, actually, we'll start with that. dr. pak, let's get your side. let's all just get a moment to think about that. what's the headline that you

want everyone to go home with or be thinking about tomorrow from this discussion. >> i guess it would be critical safety issue needs attention. >> >> sorry, i recall e i've got a long sentence. >> i would say this it would require coordinated action between all the stake holders. >> mine would be that to drive fatigued is to drive impaired, which i don't think people appreciate.

>> mine would be drowsy driving represents a much larger problem than previously recognized. and that there are a sequence of steps that need to be taken in order to address the major public health risk. >> great. there's been a lot of reference to the substance impairment issue. it's an advocacy hearing for me e mine. so the numbers for 2012, are 173,000 injuries 27,0 00 of those are life-altering. we don't have that kind of data. the other thing is the issue that is much more complex of alcohol or drug intake of what has to be done. i just want to thank this panel again. i don't know if you were noticing, but we couldn't write fast enough to cover everything. my thanks to all of you for a great closing, e. i also want to thank dr.

malloy who did just a tremendous job in creating an excellent program today. and in closing i want to thank all of our panelists and its effects on driving from a very broad range of panelists. i think future directions on how we can reduce the number of drowsy driving crashes on our roadways. before concluding i want to high light just some of the few big take away messages. one is scope. clearly underestimating the problem. and i don't think there's any question we need better data. second, what we're getting is the complexity of this issue just the diversity of causal factors here seem endless. i won't say endless, but just huge. we talk about schedules and sleep disor eders. perception is part of a long, long list. you talked about segmentation for the marketeers.

that's part of the problem here. it's just how large the causal factors are. regarding next steps, we heard about education roadway design and construction, technologies, the diagnosis in treatment disorders, organization of knowledge, implementational programs, all of this together for me the bottom line, there is no magic bullet. this is going to require a comprehensive approach with on going efforts across dayiverse statements of our society. it's clear to e that in ees's clear that in order to reduce the number, we have to think more comprehensively. this begins with sleep. sleep is foundational in biology as human being and our ability to function. drowsywoe inging habilitate

habilitate habits are deeply rooted. how many times have we heard someone, probably you, even, say something similar implying that he or she doesn't need or have time for sleep. franklin made his statement over 220 years ago and it's still juzed today. the modern version as a badge of honor, i'll sleep when i'm dead. when so many of us are troovl eaveling by road or highways, it is a deadly combination. it's taken too many lives and considerable effort to advance public awareness about the perils of drinking and driving. once it was all too common to hear one for the road. today, we're morelike likely to hear friends don't let friends drive drink. this is not the case with drowsy driving. an impairment every bit as

dangerous. the time is long overdue to address the risk of drowsy driving. to engage in a national discourse and make all of us safer on the road. we have to be awake and alert to drive and stay alive. we stand adjourned. >> friends colleagues, country men, thank you for sending me here. and let's, today, welcome all of the new numbers and all of their families to what we all know to be a truly historic day.

>> today is an important day for our country. many senators took the oath this afternoon, 13 for the first time

>> good morning. our goels at hutchins are to both improoif public understanding of fiscal and monetary policies and to be a place that gathers people from 5:00 deem ya, government and business to try to improve those policies. the hutchins center was made possible by the generosity of glen hutcheins. our topic this morning is fiscal uncertainty. we all know that projections for federal government deficits over the long run show that we're on

an unsustainable path. but it's also true that those projections have very wide confidence spans around them. the question of how that uncertainty should influence policy is quite controversial. some argue that because we have little predictability, we should instead focus on the here and now. while others argue that the future could turn out worst e worse than we expect, we should pay greater attention to issues in the long run of uncertainty. the question of fiscal uncertainty is a hugely important but whoafully underemphasized topic. and we're lucky to have some of the world's leading experts on these issues here today to discuss and debate it. our morning will proceed as follows. our first paper is on the big question of whether uncertainty means you should pay more or less attention to project a

fiscal imbalance, as by allen arabeck. the paper will be discussed first by professor charles mansky and then by noble lawyer yet peeter diamond. that panel will be expressed with much practical experience consisting of bill hokland gene spurring, former economic policy counselor and tennessee congressman jim cooper.

>> let's get started with the first paper: i'm particularly please today have alanan alanan alanback here. i still tote around my notebook from alan's class at harvard. i know what he has to say is worth paying careful attention to. so without further adieu, please welcome aaron alanaback. >> thank you very much, louise. this is the title of my paper and it's a very certain title. there's more uncertainty in the

paper as the subject. so more interpretations show a significant imbalance under current policy. now, of course that in itself is a question that one has to deal with is what is current policy. never the less there's the significant imblaens between expenditures and revenueses that leads to a very large fiscal gap. several% of gdp on an annual basis, much larger than physical adjustments were accustomed to making even as a review of large policy changes. while on the other hand projections are also very uncertain.

and, about that, there's a disagreement. and the uncertainty goes up with the horizon over which one is forecasting. just as a very model illustration of this here's one finger from the paper. taking from a cbo document in 2008 showing confidence intervals that is statistical predictions of likely outcomes as of mid fiscal year, 2008. of course, starting fiscal year 2007, when the deficit is a share of gdp was known. as you get up to the highest and lowest series are the 5% and 9d 5% competence interval. and there's also 25% and 75%. the solid like e line is what

the forecast was, the listed forecast was for the fiscal years. and then the dotted red line is what actually happened. now, i have to adjust what actually happened because the predictions were made under e under current policy. so the redline isn't what the deficits actually were, but, by cbo'sest mates after the fact the deficits would have been without changing in policy. and you can see that even in the first -- the second year of being prediblgtive fiscal year 2009, the prediction was -- the actual value was outside the 90% confidence ban. 2009 wasn't your typical year. it's reasonable. it may very well be that these confidence bans were accurate and we just had really unusual draw in 2009, as we know we did. e blaeng blank

blank

blank. >> of course, there's news about that. it's rather about how our spongss should differ as a result of there being want of uncertainty about these projections. you may find it hard to separate, but i think it's posht to do that because the arguments about one tends to spill over into the other. there's a couple issues to talk about. one of them isstein's law, which many of you know from curtstein. anything that can't go on, won't. we know that's not going to happen. some people take comfort in that. i don't understand why.

knowing that something is going to change doesn't really give you any information about how it's going to change, when it's going to change and what the con kwenss of these changes are going to be or the delays in making changes. what i'd really like to know is if we don't pay attention and wait for things to fall apart how will the outcome prepare to what we could do if we were actually designing policy. a second common response is projections behind and you can fill in your favorite number here are years. 2010 years 2005 years, one year. you can choose it.

i don't actually understand what that means. so does it mean that we have projections getting worse and worse over time saying 25 years out, with increasing uncertainty that it's getting worse and worse after time. does ignoring it say oh the problem went away after 25 years. this thing that looks bad, but uncertain, suddenly, 25 years out, no problem. if not that, then what? so what my paper does suggest is su ported by the economics, is that 23 the future is uncertain there's a case for saving as a form of self insurance, if you will. and that's an argument that comes from the way individuals should behave with their own certain prospects.

although it's more complicated, it's an argument that one could make for the government, as well. uncertainty is bad. putting resources aside so if really bad outcomes were to occur, you'd be otherwise more protected than you'd usually be. so going through several of these in the paper, i won't have time to discuss all of them but let me just deal with a few of them here. one is it's not that we're uncertain about it it's just

that we don't know the future. this is unfortunate. obviously u we wish otherwise. but that's life. it's not something we should ignore. it's hard to come up with a coherent argument of why we should ignore things for which we're very uncertain. another common argument here is that people will be better off in the future so that they can absorb greater fiscal burdens than those being imposed on people today. the first thing to say about this is this is really an argument about having to deal with projected physical analysis. again, noting this distinction. that is if we knew for sure that taxes would have to go up or benefits would have to go down for people in the future, that would be an argument one might support based on the greater one

being in the future. it's not really the argument about how to deal separately or rendition to uncertainty. indeed, there are arguments suggesting that the uncertainty exacerbates fiscal problems for the future. in particular, if we get a series of negative events that mean that taxes don't gist have to go up as we expect them to go up but have to go up a lot more, we could have very serious economic damage as a result or some sort of fiscal crisis if we find ourselves able to collect the tax revenue we need to. i think that actually pushes us toward dealing with the problem now sitly than projeblgtsing it now, as this would. second we should wait until we have a better idea about the future. well, this may be consistent with a view of the world in which there's sort of a certain amount of ene uncertainty out there. and we'll resolve it. as time goes by. well, the first thing to say about this, there's always more uncertainty coming.

that is we may resolve ef some issues, but other new ones about which are uncertain will rise. and, you know, we wait. we're not going to make uncertainty go away. but we will restrict our options for dealing with it. second, there may be cases in which we do expect a resolution of information 06 uncertainty about certain things. we may discover where the provisions will actually do so. it may cause further delay because we want to learn more

about the most efficient way of delivering of health care. but it doesn't mean that we shouldn't put resources aside to make it easier to deal with the fiscal uncertainty that we face. there are certain costs leaving aside the political costs to make it happen, there across from the economy. so we might want it to act more

gradually or wait you believe till we really feel we need to. but, on the other hand, when we do act we should act more forcefully knowing that we're not going to have an opportunity to do it in the near future. and finally and this is something that david caymen talks about in his paper, we can put in place some automatic responses. if we can be fairly confident about what those responses should be as events unfold. an example of this might be and i'm not advocating this, indexing social security, the age of retirement under social security to life expectancy. that's been proposed by many people. it's not something we have. it would certainly make it easier to keep the social security system solvent. on the other hand that might not be a good response for life expectancy. again, it would be much more

increased to the retirement age than others and maybe with a different response. but the key is that the automatic policy responses that we build in ought to be things where we could say if an adverse effect occurs, we have events of what that's going to look like and this is the response that we're going to move on. with greater uncertainty projections are more susceptible to political influence. the story is the agency doing the forecasting doesn't have a strong argument about what might happen, then the political influences might be more able to change what the forecast is. hard to defend against this influence.

i think ignoring information is a very bad choice. it's certainly one we should make only as a last resort. so, to summarize my comments, and my paper u which as i say, goes into more detail than any of these points i've made. uncertainty will always turn out to be wrong in some sense. it's going to make life more difficult for us because we're going to have to make adjustments and we know we're not going to get it right. we know we're not going to get i right except for coincidence. but that counted e doesn't mean that ignoring uncertainty is the right thing to do. we can't make it go away by ignoring it. it's still there. it's better to form late a response than simply to ignore it and wait for things to happen. thank you.

thank you. very happy that the hutchins center has organized this event. my remarks are going to follow on alan's, in the most broad principles that we should be facing up to uncertainty, rather than ignoring it. i've been trying to get that message across for quite awhile in my own work. i didn't write a paper specifically for this event. but here's a couple of sources at the bottom, if you're interested, of book, public

policy and an uncertain world published a year ago, and then a article that i'll talk about a little bit today on communicating uncertainty and official economic statistics that will be out in the journal of economic literature, i think, next september. and that's on my web page if anyone wants to take a look at that. let me take a moment on the general themes of my work. they are foremost that society should face up to the uncertainty that attend policy formation. and i've been quite critical of various practices and policy analysis that hide uncertainty rather than facing up to it. my background is an economic rather than public economics. i'm very careful about how we make empirical inferences, and often the way we make empirical inferences is taking whatever data is available and adding whatever assumptions are needed to draw strong conclusions. and that may not seem a red

button for most people, but for me that's what really gets me riled up. and so what i've been arguing is that a credible policy analysis would explicitly express the limits to knowledge. i try to show how that might be done in my technical work, and on the broad notion, is to study how policymakers can reasonably, i won't say optimally, because that's going to far. i think that's pushing things. but reasonably is as far as i would go, make decisions in an uncertain world. and i think that's what alan's paper was focused on. now, if we're going to face up to uncertainty. alan was focusing on projections to the future. but we have to face up to uncertainty, even of other things that we think we know about the economy today. so, i want to talk a bit about communicating uncertainty in official statistics. so, we have all kinds of statistics that summarize the state of the economy that are reported as point estimates. these could be unemployment rates, you know, growth in employment, gdp growth, household income statistics, and

so on. that the federal statistical agencies report. and if you look at the news releases that come out monthly or quarterly, you find out that there's very little mention of error in any of these statistics. if you dig down into the technical publications, and i've done this, from the census bureau, the bls, and other agencies, you'll find verbal acknowledgment regularly that estimates are subject to sampling and nonsampling error. you'll find a little bit of trying to measure actual sampling error by confidence integrals and standard errors. but most of the error is really nonsampling error and you won't find any quantification of the nonsampling error. so reporting official statistics as point estimates manifests a tendency of what i call policy analysis to project incredible servitude. you make as if things are certain but there's really not much credibility behind it. and agencies do not justify the

way that they produce point estimates. let me give just three examples that i think are important. one, if you look at gdp estimates that come out quarterly, those of you who know, the bureau of economic analysis reported advanced estimate, and the second estimate, a third estimate and then at the end of the year, and after five years there's a continuous process of revision of the estimates. and the revisions matter a lot. you can think from the first estimate that gdp's gone up 1%. next month you go oops it went down 1.5%. the estimates that come out are not accompanied by any error measures. the way the bea does it is by doing trend extrapolations when the data is incomplete and replacing them with real data when it comes in. it's a process that takes a lot of time. the bank of england actually, in the british context, actually puts in its -- does a fan chart that puts error bounds around covering gdp estimates.

and i think that's something that we could do here. a second source is in sample surveys for anyone who uses sample surveys you know there's missing data. people don't -- are interviewed or they refuse to answer questions in the current population survey the basis for our household income statistics is a huge amount of missing data on household incomes. upwards of 40%. and you'd never see that from looking at the press releases that are put out, because the census bureau imputes everything. and you got to dig down and teach econometrics to know that these imputations may not have much value. there's big potential problems there.

a third, i could go on at great length about any of these issues, but a third that drives not just me crazy but macro economists regularly crazy is the use of seasonal adjustment. if you find out that the unemployment rate went down last month relative to two months ago, did it really go down or is that because of the seasonal adjustment formula, the x-12 or x-13 from the census bureau through its auto regressive moving average process made it look like it went down. now in this building i don't have to say much more about this. jonathan wright, from hopkins, wrote a very nice paper on this, came out as a brookings paper just a year ago. so we have all of these issues. agencies could use established principles to report sampling error in statistics. it's more challenging to measure nonsampling error. but i think that good faith efforts would be more informative than reporting official statistics. even if it's hard it's better to do it than not to do it. why is it important to communicate uncertainty. governments and private entities use official statistics when making decisions. the quality of decision making may suffer if decision makers incorrectly believe the statistics to be accurate. or if they think they're sophisticated and they have to guess at what the error margins should be but they don't really know what the magnitude of the errors are.

i think it would be better if the agencies, the statistical agencies would communicate the uncertainty then we would have a better understanding of what information is actually available about the economy. now let me end the rest of my time, let me move forward, and talk about projections. i've written about this before. i warned doug elmendorf i would raise this question again about cbo scoring practices. the congressional budget act that established the cbo has been inerpreted as mandating the cbo to provide point predictions or scores of the budgetary impact of legislation. the scores are conveyed to leaders of congress, and they're not accompanied by measures of unternty. there are various things the cbo does that does express uncertainty but not the scores. i -- one notable example that

i've used on a case study in my book was the scoring of the affordable care act, and it's a quote from the letter that doug sent to speaker pelosi back in 2010, the cbo and jct estimating that both pieces of legislation -- people will remember there were two pieces of legislation. would produce a net reduction of $138 billion in the debt. so the question is what does that $138 billion reduction really mean? plus or minus $5 billion? plus or minus $50 billion or whatever? doug will remember, douglas holtz-eakin wrote in the new york times that the real number was going to add to the deficit by $562 billion. his estimate and the cbo estimate were off by $700 billion. so what i'd argue is that the cbo should express uncertainty in scoring. and i have to say -- and this is controversial. okay. i give a talk at the cbo a couple years ago about this. cbo has established an admirable reputation for impartiality. maybe it's best to leave it alone and just continue things as they are. i worry, though, and maybe this

is just my being an academic from the outside who doesn't understand washington, but i argued that there's been a social contract to accept cbo estimates. i worry that social contract is going to break down at some point, someone will dig in, in congress or the media or dig in and find some estimate from the cbo they don't like, ask how the sausage was put together, and then say, well, you know, who knows. and then this cbo's reputation will go. i would rather the cbo face up to this, and provide uncertainty in its scoring measures. very simple way to do this might be to provide integral forecast and upper and lower bound. it could be a probability. we don't have to get into those details. now the question is this. the right question to end this room, can congress cope with uncertainty? because when i've talked about this to academic audiences everyone says to me of course the cbo should, you know, should be transparent about the uncertainty in the scores. when i talk to people around

washington they tend to be skeptical. there are two reactions that i've received. some people assert that members of congress are psychologically or cog knitively unable to deal with uncertainty. i've heard that many times. i won't ascribe it to any particular individual but i've heard it many times. some give a game theory argument that congressional decision making is a noncooperative game and expressing uncertainty will make things worse. there was one brookings economist with whom i had a fairly uncomfortable e-mail exchange about that a few years ago. i'll end up with three questions. because i can't answer whether congress can cope with uncertainty. so i want to just end up with these three questions. first, how do users of official statistics and cbo scores actually interpret them right now? second, how would transparent communication of uncertainty affect policymaking? and then third, and this is what this latter half of alan's paper

was about, is what would constitute normatively reasonable fiscal policy in an uncertain world? thank you. >> alan has given us, first of all, a clear picture of uncertainty out there, and it is large. i want to do a few things around it. first i want to, no coordination here, quote chuck manski. that there is a important political question of how different ways of conveying uncertainty would impact and that matters. before i knew what chuck -- before i even paid attention who else was on the panel. i thought this really belongs. and what's important to keep in mind here is what we really want around policies is full blown benefit cost so we can think

about why we like it. why we don't like it. first of all, there's no way that can be done by a neutral agency. because if nothing else it's going to involve way to put on different concerns, way to put on different parts of the income distribution. so the issue is, what can be done by an agency like cbo, which will help with the process? and i think we need to answer the questions that chuck just posed for getting on with that. and conveying uncertainty should always be there. but the question is how, and how that fits in. so i want to focus on social

security for the odd reason i know a little bit more about it than the overall budget.e & so, the office of the actuary does three projections. and that's conveying uncertainty. it's often criticized because there's no easy way to hang probabilities on those two outsiders that are meant to be unlikely but no ease he is way of doing it. and there's also a sarcastic projection which i think has the unfortunate effect, again, chuck mentioned it, this is leaving out uncertainty about the model you're using. it's leaving out uncertainty about the nonstationarity of the time series you're using to set up the monte carlo. i think it's really important to have both. just because they're going to communicate, and what we can hope to do is communicate. so, let me go to the three issues that are -- everybody is talking about.

i didn't invent these, around the uncertain projections. automatic adjustments. legislation for future implementation, and additional savings of what i'll call alan's theorem, in response to increased uncertainty. i want to say i love automatic adjustments. i think it's important to have legislation for future implementation. and i think we need a lot more theory work done before we accept allen's theorem. this is not to say i think it's wrong. but to say we need some serious studies. and let me just run through some of this. automatic adjustments, our security has some of them, adjustment built prices, adjustments in wages. that's it. you look abroad, sweden has adjustments for mortality rates for life expectancy, both for initial benefits, and the increase in benefits to delay claiming. germany has a benefit adjustment

for the old age dependency rate. and sweden has an adjustment which they made an absolute mess of. if the projection of solvency gets out of line. jumping ahead, having read the abstract of kamin's presentation, i think trying to deal with solvency projections automatically would be hard, and not a good idea. doing something for mortality, as peter orszag and i proposed in our book i think would be a good idea. but again alan's example indicated the importance of being careful for that. and let me just throw out the point that cbo, in its projections, assumes all the benefits will be paid, and there will be no increase in revenues for covering that.

perfectly sensible, central projection. it's also the projection in the alternative fiscal scenario. i think the probability i would hang on that outcome is very close to zero. it's hard to imagine its debt, finance, all payment of benefits with no change in revenues since two real histories of it in '77 and '83 show benefit cuts to be part of the story and revenue increases to be part of the story. so just to move on then to current legislation, future things. some of them have no credibility and don't affect private behavior. and turn out often not to happen. in social security the history is rather different. from the beginning of social security, up to 1990, there was always a future tax rate increase that had been legislated. none of them were ever repealed. some of them were delayed.

some of them were accelerated. it clearly had an impact on the ability of congress to deal with that. and the increase in the age for full benefits, which was voted in '83, has had no serious points. so trying to figure out when you can do this in a way that leads to better policy seems to me to be an extremely important point. and now let me move on to allen's theorem. that's the theorem from him. but he draws on a model of individual behavior for it. which makes the point that there are conditions unique to that. and secondly, he makes the point that with uncertain rates of return, it becomes more complicated. and here's i hope i'm not being unfair, 100% of the logic behind, you go from the individual to government. i'm sure there's more in alan's head than was in the paper, particularly what was in the draft i went to. so let me talk about what missing theory there is. and first i draw on a paper by, of all people, alan and kevin which made the assumption that when you had legislation that triggered a delay till the next

legislation, i presume, i didn't actually read the whole paper, that congress was acting in a optimal, consistent way, around that one political constraint. but i think the point to recognize here, unlike the models of individual behavior, is we don't expect continuous adjustment, but periodic adjustment. the second element is what kind of adjustment do we get? i don't think we want to be modelling, congress is always been consistently optimizing over all in a way we approve of. and i look around whether anything's on individual behavior, one might draw on, and this paper has behavioral misbehavior when you're making the decision, and uncertainty about what will happen if you try to do ahead of time and analyzes how to set the budget constraint for yourself to trade these off. i think that's an interesting mind-set for going forward. and now i want to turn to dead weight burdens of taxation which are part of alan's presentation. and my hobby force is, if you ever say the word dead weight burden, you need to say something about the income distribution changes that are accompanying the particular level of dead weight burden from the particular tax policy you

this paper has behavioral misbehavior when you're making the decision, and uncertainty about what will happen if you try to do ahead of time and analyzes how to set the budget constraint for yourself to trade these off. i think that's an interesting mind-set for going forward. and now i want to turn to dead weight burdens of taxation which are part of alan's presentation. and my hobby force is, if you ever say the word dead weight burden, you need to say something about the income distribution changes that are accompanying the particular level of dead weight burden from the particular tax policy you are looking at. obviously, if we had lump sum taxes, we could have no dead weight burdens. but we can't for asymmetric

information. if we had a small enough budget, and a good enough population so we could have a uniform head tax, to cover all of the budget, then we could have no dead weight burdens. but we might not like that. we might choose to have dead weight burdens in order to have a different tax structure for better income distribution. in that case, the presence of dead weight burdens is a sign that the policy is better, not worse. and emmanuel saez in his thesis asked the question, if you can't finance the government with a bold tax and you don't want to pay attention to income distribution, what do you do? and it turns out you get a particular optimal tax model, and he goes ahead and solves it, and my thinking is we need, on the policy adjustment, rather than the whole budget argument, an analysis of the minimum dead

weight burden. that counts as dead weight burden. and if the actual policy is anything different from that, that's making the policy better, not worse, and i just want to remind you of two standard welfare theorems in the context of public finance. the absence of distorting taxes when you have income distribution concerns, and some other conditions is a sign that you're not optimizing, and even if it's just individual uncertainty that you're dealing with, with asymmetric information. again the absence of distorting taxes is evidence that you don't have the social welfare outlook. thank you.

[ applause ] >> thank you very much for that. score you high on two counts. one is clarity, and the second is brevity. as we know they don't always coincide at brookings so appreciate that. alan, i want to ask you about something that peter raised. and it basically goes, as i understand your argument, you're saying that we should do enough today to put the federal government on a sustainable fiscal course that involves changes to taxes and benefits. so whatever you think we ought to be aiming for we should do that. and then we should tighten our belts more to account for the fact that there's a lot of uncertainty about whether --

>> yes, that's basically it. >> and, that sounds like we have to go through a lot of pain that may not prove to be necessary. and you -- and the reason we should do that is? >> the reason we should do that is that -- first of all, say life is more complicated than that. but, starting from that, you know, basically what i'm saying, if things -- there's a lot of uncertainty. and we have a series of adverse events which cause things to be within the predicted change but a lot worse than the baseline estimate. then if we haven't taken forceful action, we're going to have a disaster. economic disaster. not just a, you know, some people paying higher taxes, but a real economic disaster. >> and it's not enough to build

in indexing or triggers. it's we have to save more now? >> well, you could build in -- you could build in indexes or triggers. the problem is that that overcomes the policy problem. you're enacted it. you won't have to enact it. but for example suppose the trigger is marginal income tax rates go up when there's a bigger revenue shortfall. well, it only happens after the shortfall, or when the short -- there's an incipient shortfall that's still going to give you the same path of taxes that you would have done if you were simply responding, and that could still give you, you know, very, very high marginal tax rates in the future. or very, very low socioinsurance benefits. so it overcomes the political problem, and perhaps avoids a fiscal crisis from the -- in an inability to act, but it doesn't change the economic costs or the adverse distributional effects. >> you think he's right or wrong? >> both.

how's that? >> good. >> the point here is he was describing a particular way of responding to triggers that are there. and first of all, there are other ways to construct triggers that are forward looking. and secondly, one of the problems we have with some of these things is congress will have powerful incentives to undo them rather than do more of them. but we also have the ability to. so i think until we get usable, useful picture of the interaction of position and political action, it's hard to jump in spite of that. i don't think they're saving more now is necessarily part of the optimum. it may be. but, inallin''s sense of saving in terms of the cap, i find the

argument more likely to have the conditions. >> you want to weigh in on that? >> i think -- let me add to what peter was saying. i think actually, it's uncertain. which goes with this morning. alan began with a -- alan's presentation was, people like peter and myself, allen never feel comfortable unless we see a well worked out model. of course that wasn't in this paper but as peter alluded to, i don't think it's there in the literature, even in alan's earlier work in 2007. when i was in graduate school and i learned that savings

arguments, the effect of uncertainty on them depend on the third derivative of the side the political economy of how peter was reacting to and if this were a purely social reacting problem, this is a new problem. so what allen is doing is doing a disservice by sticking the neck out by making a strong proposal which i hope would lead to the research that would get underneath us. >> and you have not spent all of your life in academia and you have spent some time on the joint tax committee a and do you really think that members of congress even if we say, the median and above members of congress can actually distinguish between what's a projection and wa h's the uncertainty around that projection? are with we being a little bit

naive here if we think that is a little bit possible? >> no, i don't ti that we are being naive. i think that obviously that some of the members are more expert this this than other, and members of the budget committee that are likely to understand it better than people who are not. and i also think that while they may not understand the terms of understanding the statistics in terms of the way that we do, you know, the characterization of what the problem is of posed by uncertainty, i think that can be done in a way that is comprehensive to both educated people, and the experienced people who are like members of congress. i do, and peter mentioned the ishsue of the political problem, and doi see the political problem in for example trying to puts a side the resource pors the future. this is not the problem that comes up with us. it comes up when the countries

have a temporary in flows are from the natural resources not disappearing in the future, and understand that they need to put the resources a aside and have difficult doing so, and so i understand that there is a political problem when our theory says to put the resources aside, and you know but yet the resources are there and that it is tempting to spend them. >> let's tourn the audience, because we are have time or the that, and there is a microphone going around. p if you have a question, it would be helpful if you tell us, a, who you are, and b, follow the panel's example of brevity. >> my name is luka bocelli from the hill panel report, and i hope i have not misread this because i only read one page, but you said ha the sources of uncertainty over the next year or the time horizon and the past cycles is productivity growth, and interest on the debt, and

the health karcare cost rising, and if you said that those are the sources of uncertainty, why not push them in the positive resources, and so which one for example? >> i don't think i understand what you mean by pushing them in the positive direction. >> basically, well, first of all, it is, it's kind of hard to come up with anything that has a substantial effect on any of the factors in terms of the policy. and second i'm writing really here about uncertainty, and not about the factors and again, i want to separate what we should do about the projected imbalances than what we should do about the uncertainty of the projected imbalances, and than what we were to do with the projected growth and that would

mean a smaller job ahead of us, and it does not tell us how we should be responding to uncertainty. >> in the back. and right there. yeah. >> james, hi. a question for allan, also. when you use the word savings, there is always an interesting relationship of savings and investments and models, and if the savings were just people sitting on the money, it would not only just reduce the uncertainty, but it is the savings would return to investments, and in that case, why wouldn't the government make the investments. >> well, when i talk about the saving and the paper, i am thinking about the government retiring debt, and the simplest thing i am tauklking about and i'm not sure what this issue is when it came up. >> and thanks. >> i agree with allan's basic conclusion that we should be doing things in advance. the question of why save? well, one reason that we should be thinking of saving is that we have seen the size of the

national debt double relative to gdp in the last decade or so. and so from that point of view alone given the risks associated with the large debt, paying some of it down, reducing the future interest costs and the dead weight losses, associated with raising the taxes to service that debt, it all seems to be a good thing. another thing that sis particularly thinking about social security and medicare, and yes, h there is uncertainty about the future and when the time came you cannot adjust the future of retirees quickly both as a political matter and in fairs on the the individuals. you have to give them time to the adjust, so maybe we gave too much time to the social security reforms of 1983 which we are still phasing in 30-plus years later, but you can't do it overnight.

so that is another reason to re respond to uncertainty by looking ahead and making the adjustments. i like alan's suggestion about indexing to life expectancy. and he both made the case for it, and also raised the concern that not everybody is going to enjoy the same increases in life expectancy expectancy. i think that we know the statistical evidence is that lower income people have not enjoyed the same increase in life expectancy over the last decades that higher income people did, but that is remniable, and not individual by individual but income group by income group because with social security, we have the records of lifetime earnings so we can make the age of full

benefit depend upon the average income during the individual's working life. >> two questions really. you talked about uncertainty, but there is often a useful distinction between uncertainty which is stuff that you can't price, and risk, this thing that has probabilities on it and so it would be useful to make those distinctions to make the things that you are making about. and my second question is from monetary policy. there is brainnard uncertainty, and that is that you should move e slowly and see what the effects are when you are uncertain about the effects. is there an analogy here or the political stuff that you would see is very, very hard, and i wonder if there is any analogy here.

>> on the first one i talk briefly about the distinction of the risk and the uncertainty suggesting that perhaps ha is what underlies the people's arguments about being not just ig ignoring the uncertainty and the niaivete of the unun ununcertainty, and while i see that, but i looked in the literature and there sis a small literature about the ambiguity aversions and not even uncertainty, but there is no argument, and nowhere in the literature is there an argument to ignoring uncertainty, and in fact, it tends to be analogous, but it is a fairly thin literature. and the second point i guess that i would and i had not been thinking about the brainerd paper, but the arguments about the health care reform have that flavor. if we are not sure what measures we should be taking, then we ought to take precaution there

but that does not mean that maybe, so maybe the responses should be more of the budgetary variety and less of the structural variety if with we don't know what the right way is. >> and time to look a little bit to the -- >> well, to the say that we have a medicare projected deficit, and we don't know how big it is going to be and we don't know how various the reform measures are are going to change it, and so we have to be careful in undertaking the measures, and it does not mean that we can't raise the medicare premiums or do anything else to improve the funding of the paper peter. >> let me look at thes aionals of the brainerd paper and it was turned down and i never followed up on it. he assumes something that you know perfectly, and there is a parameter of changes from that where you have uncertainty. and so you go toward what you know perfectly. the problem that alan has posed for us that there is nothing

that we know certainly. so i don't think that brainerd analysis holds or the monetary policy or fiscal policy either. >> and also on brainerd, take it as another example as peter was saying precisely what is uncertain matters a lot. you can get a different result so you have to have the model. on the biggerr issue about 99 uncertainty or ambiguity, and what we need to talk about is the view of the nation and what it should be regarding uncertainty. once you move away from the expected utility framework we have to ask if the government should be reverse averse, and once you move away are from am ambiguity ambiguity, and this is doing to show nup