mployed. that's not really an indicator of the economy going up. that's just people falling through the cracks and being forgot about. if you look at it from that perspective, the rate of unemployment in this country is probably 10 point something higher than what it's actually beingest mated as as far as the figures of what they're showing. >> continue to let us know what you think about the programs you're watching. call us at 202-626-3400. e-mail us at comments@cspan.org. join the c-span conversation like us on facebook, follow us on twitter. >> next, a look at the budget scoring process with economists and tax policy experts. they disz cusscuss the government's procedure for drafting a budget and the steps taken to make budget figures and other information easy to access. this was hosted by the brookings institution.

it's just over two hours. >> for those of us who have been in washington for a while feels a little bit like this person will say this and then this person will say the predictable thing and then marty sul van will say this and bob mcintyre will say that. so we'll probably have a little of that today. but our goal today was to move a little bit beyond the conversation about whether dynamic scoring is either the

best thing to ever happen to washington or the worst thing to ever happen to washington. and, instead, to focus a little more on what is it actually mean and how do you do it. what we're going to focus on today is skpaktly how it would be done. what assumptions you'd have to make. what are the challenges that the jct and the cbo staff face in

doing it. how can it be best presented so that members of congress and the rest of us have a clue as to what the analysis does. so this is a bit of an experiment. and, moving beyond the basic pissing context in washington to look at a little bit of facts and information. let's look at how we're going to do this today. i'm going to start i'm going to moderate a conversation about the big picture len is now the director of the tax policy center. and he also was the -- deputy assistant secretary for tax analysis in the treasury in the late clinton years and has also worked at cbo. doug holtz heaken was the director of cbo from february,2003 to december, 2005. so both of them have grappled with these issues, firsthand. following that, we're going to turn to something we're very

pleased to have a presentation by the staff of the joint tax committee on how they actually look at how they do this and the mechanics, followed by a discussion that my colleague, bill gail, who's one of the co-directors, who is the co-director will moderate. and then, following that, and you can't leave before the last event, we're going to talk -- i'm going to talk with donald maren and steve mcmillen about the application of dynamic scoring to bills other than tax bills, which is also required in certain circumstances by the new house rool. so, with that let me invite len and doug to come up and we'll get started. >> i should mention if we can keep this on time, we're going to have questions after each of the three panels.

and, in case you didn't notice there's tv cameras in the back. one is for our web cast, there's people watching this live and it will be archive on our web site. and the other is c-span. and we also have on our web site a little annotated bibliography that my colleague made that has references to the official documents of the act and cbo on this as well as things that have been written by each of the people who we have as panelists today. thank you for being here. doug, i want to start with you. i wonder if you could describe to us a little what was going through your mind when you were at cbo and you decided it was appropriate to do a dynamic skeer, that is to factor in the economic effects of the president's budget and what did you learn from the exercise? >> so it started actually prior to my arrival at cbo. it became pretty clear during the selection process that this

was spg that was important to the members both sides of the aisle. both sides of the congress. as a matter of principles, you want to look br e before and after and everything in between. it seemed to me that the president's budget was the ideal place to do it because it's the most fully specified single set of policy initiatives. so that seemed like the right thing to do. it also gave you the right comparison. every president's budget is dynamically scored. if you read them, this president, bush, clinton, it always says these numbers are contingent on the president's plan.

so that was comparison in my view. that was something that was pretty uncharted. i don't think you should think of those as anything other than guidance. you have to use the models for some help. but you never have anything in the model that the congress or the president imposes. those things are incredibly clever. how do you present the results? how do you communicate on them? there, i thought, you know we got a really, really solid f. at that time there were some tax cuts that had some fairly straight forward supply growths.

and then there was the medicare prescription drug bill. i thought the con would conclude that, gee it's not a good idea to offset growth policy with this big, new spending program instead of the we did it wrong. that was it. >> so it seems obvious that congress should consider the macroeconomic effects when it's considering a major piece of tax legislation or spending legislation. so one of the pros and cons that are folding the official score, whether it's the particular piece of legislation that will come ply with some reconciliation or some other target.

the con is that there's so many cases that actually have no idea. a lot of the cases don't even know the sign of the effect. there's the models, themselves are these kind of stylize edd general equilibrium models which means basically, they're trying to account in a very kind of aggregate way how, say capital and labor affect the economy and how taxes on capital labor affect investment and work effort. those are always bassed on estimate e mats that are themselves, very uncertain.

we have data compiled by consumer finances on how much assets, how much capital gains have in their portfolios. we can calculate ruffly what the revenue might be. first of all, we know it would raise revenue at least before -- in that case i think it would raise revenue certainly before including any kind of macroeconomic effects. and then if you were to apply dynamic scoring to that that issue, they're in the models. they'll say we're cutting the tax rate -- we're raising the tax rate on capital. there will be less saving, less investment. that would hurt economic growth.

in the real world, what we're doing is we're raising the tax rate on capital. but we're also removing the single biggest loophole in the individual income tax which results in a huge amount of unproductive tax sheltering. we don't have a good idea of how important that is. in this case, there are people who believe really, really fervently that is going to be one answer or the other answer that there will be an enormous pressure on estimators in that case. they did an analysis after it was done. that was a policy which all 06

us working on tax reform are convinced that it's good for growth. the net effect was probably zero but it's really hard to tell. >> presume fwli, that gives jct the right and responseblety to say we don't know. if we had to wait until we have perfect knowledge before we use these things,that seems like an unreasonable standard.

i just looked down at my notes. political restraints, this is a good thing. this is removing the con strant on theest maters that my concern is exactly what you're talking about. that they'll actually be pushed -- that congress would be mad if they say we don't know. i'm very sympathetic because we're just saying this is good for growth or bad for growth. but the problem is the answer should be no if we don't no. >> do we know enough about this? >> yeah, we know enough about this as just as much as efg educational that we score. >> i think we know a lit less. >> i'm not sure that's true. honestly. we know that the precise answer

of zero is exactly wrong. there's no virtue to being exactly wrong. by knowing that there might be an impact and saying we'll ignore it. so i think sort of doing this makes sense. remember, this bill would probably not get dynamically scored. it's not big enough, the one you're worried about. this is only done in the house rule and should only be done for a major piece of legislation at large predictable impacts. so most pieces of legislation, it's not going to apply to. >> right. >> this is a side show most of the time. it's not going to happen very often.

you do want to look at it. it is always a scoring issue that, to look in the research literature and sort of see what the consensus tells you. and, in many cases, you don't know much. i have a long list in conventional scoring that are comparably uncertain to what we're facing here or worse, quite frankly. and in those circumstances there's a lot of zero that is get put in there. if you think back to the aca, the affordable care act, advocates desperately want to see cbo score the prevent tich measures of saving money and they didn't. they were mad at it. i don't think there's anything that changes political pressure here. the staffs are not slaves to these models. it's not like they have to say the model did this. scoring is a judgment science and a judgemented art at the same time. we should hire high quality

people at the jct and the cbo and we co. do. we should respected their judgemented and they should describe the way they arrive at the bottom line. this is not some slavish policy. that's not the way it works. >> but i guess i disz agree with the idea that this is just like doing other scores. there was no financial product in nature that gave seniors their outpatient prescription drugs. zero. >> you actually did have data unless they spent on prescription drugs. >> sure. but the point is when it matters the most you are in the end making some judgements. it's not like you're changing something you've done a thousand times before.

>> typically, it involves a calculation on some scoring. you've got a viert of different models you can use to produce wildly different estimates. in a lot of cases you don't have any good basis for the parameterest mats in those mot e models. i'm not saying that we shouldn't try to do this. i'm just saying it's a lot harder than most of the estimates at congress. >> i can't imagine all of the things they had toest mate, that would have made it harder.

>> so doug, you said that you thought you communicate edd the dynamic score poorly. >> i think there's a big advantage to putting it into the formal score. when it's advisory and we have the range of estimates coming out of a range of model,you're bassically trying to give a master's level graduate course in macroeconomics in three minutes in front of committees and that's just not feasible. if you put it in the score, you do have to come up with a number. and they may not like that number. but jowl do have to come up with a number. it now matters. it doesn't matter when it's advisory. and now they're going to want to know about the number. and everyone out there is going to want the say something about that number. they might not like it. that generates aditsz digsal research on the issues you care about.

>> there's an idea in how much the macroeconomic events are changing. it's really important to be transparent about the chants through which the growth effects are likely to occur in the model. this is a case where you really are telling a story. and there are alter nat stories to be consistent with.

i would follow the model of capitol weather gain. they put out a forecast. and their forecast includes in their discussion the level of confidence, low to medium, medium to high. there's some e some things on which everyone would agree that this would be good for growth. at least we're sure something is going on. it's important to be able to do that. one advantage of doing that is i think it would -- it might be a chance for the scorers to tell congress the deficits particularly in the context of an economy with full employment and a surplus bonus. that would be another level of hand waving. but i think it would kind of come out of some models.

the other thing, doug talked about researchers out there. it would be really nice if congress would actually try to put some money into research. >> but the problem is you know, you can't just lay out a little model, estimate aggression and, you know, find the estimate for how to you know taxing capital gains at death affect the economy. and finding a way to encourage academics to try to answer real world macro 5:00 no, ma'amic questions as opposed to the kind that lead to nobel prizes would be really helpful. >> i'm not sure -- this is a different conversation chlts but i'm not sure we're in a great place if it's going to take congress to encourage -- >> i was just talking about money.

>> if i propose a tax cut on investment, you and i would say that would be huge. if i get the bill on my deszing and i'm at jct, do i have to decide? are they going to pay for this somehow? or do i have to assume that there's some negative affects ten years when we blow a hole in the deficit. >> so again, the key in scoring is just to recognize it's not forecast. >> but don't you have to make the forecast about what the deaf sit's impact is? >> in scoring, you do that. you want to treat all bills the same.

this is a big part in status scoring. >> in other words, they have to decide if they're going to decide whether we're going to have a deaf sit of -- >> yeah. and use the same procedures for all pieces of legislature. that's the key. >> raise your hand if you have a question so mike can find you. >> what happens with the deaf sits coming out of policies is really important. if you think of the consequences of the 2003 tax it cans which you said are pro-growth, president obama elected and had high-income surtaxes as a way to reduce the deficit pressure, the net effect maybe even in these macromodels, presumably, would be negative.

policymakers have been told a million times by cbo. but the job is to provide budget windows that they have elected. you can't cure all problems and a great mistake is for the staff to get this idea that we can somehow trip them into doing the right thing if we just show them the right numbers. that is a terrible place to go. >> an abser vags and a question. if there is a tax cut or a spending increase eventually, there has to be payment for it even for something on the other side of the ledger or through interest payments, the present value which is identical to the shift.

if you are using a ten-year window you don't want to be exactly wrong. >> we actually did this. if way you do it is outside the window you have an offsetting policy and you always use the same one. >> but it has to be incorporated into the estimated effect of the policy. >> no question about that. >> then the 10-year window is irrelevant. the question i have is i'd like to ask a practical policy that questions the threshold test. let's put together a combination of constructive realization at death, the president's proposal. and a restoration of the estate tax as it existed before the

2001 legislation. so a much lower kpechgs at a higher rate. i think in combination, that revenue would be sufficient to pass the threshold. all of which is use edd to cut corporate tax rates. do you have any inkling as to what that would be on economic growth. >> i'm not sure i understand the policy. >> the policy is quiet clear. it's an increase in taxation at the time people die and a reduction in the else state tax exemption and an increase in the else state tax rate quite large so that arguably, if there are effects on behavior, these changes would be large enough to affect behavior. i'm not sure what the sign of that policy would be all by itself, but i wanted to combine

it with using all the revenue to cut corporate tax rates. so that it has an impact budget on the floor. can you suggest what the siend of it would be to affect change. >> let's say there's a huge change in the effect on growth. >> my instinct is that's going to be positive for growth. and i'd be happy to work stlu the numbers. >> that's what the staff would have to go through.

>> net or gross? >> gross. >> the net effect is zero, but it's big enough to trigger the role. if jowl shift around snuff money, it's the gross, not the net. if the pluses or the minuses add up to a big enough swing. we'll get to this later. >> is there a regulation? >> wait for the mic, henry. be brief. >> the answer has to take into account open economy consequences. exchange rate effects. you may think there's uncertainty about domestic macromodels. the uncertainty with respect to open economy models is vast. and it depends on the response

of other countries. so i think it's on very shaky ground. >> there's an estimate on corporate taxes, as well. if you don't think about what the rest of the world is going to do. your point is it's even more complicated. >> i think we have time for another question, if there is one. there's a woman right here and then we'll go. and we'll have time for questions at the end. so you'll still have a shot. >> laura bresz e blessing. i have a very basic question. the basic question may be reiterating yours from earlier. giving that since the mid '90s, the republicans have been trying to get dynamic swearing in. and it's always been an option to have it in addition to the projections provided by the cbo. are we going to use dynamic scoring instead of the magnitude we discussed rktsz instead of

the regular scoring? or are there going to be cases where we just literally can't tell. i don't know if it's always an add-on or always a substitution or what. >> i've read critical remarks of yours that have been critical of -- talking about the difficulties with using a particularly in debt when we're, you know the deficit is going up or in debt situations. and i was hoping you could provide whatever nuance i'm missing because id eve been following dynamic scoring, but not your specific remarks. >> so i think on the basic question, i'll say what i know and then we can ask the jct people later. basically, the rule says if the

bill is large enough, score in the event required for the big bills in the house. >> and i would supplement that with the suggestion that you provide both the dynamic score, which is the official score but, also the information with the static is perfectly sensible. so you eel know. with the issue of debt it does have to balance in the long term. that's not "easy." the point is there are lots of things that are going to have to be sddsed on operational things.

what do you do without your deficits? have an offset policy somewhere out there. do it the same for every fees of legislation. these are any knowledges that i think are genuinely real and have to be dealt with. but they're not insurmountable and don't disqualify from dynamic scoring. >> i have a quick response before we turn it over to the next panel. >> i would agree with doug, comment i kpept i would do the deficit offset in year 5. >> okay. bill is going to introduce from the next panel. so thank you very much, guys. >> we are dlieblgted to have members of the jct staff here.

in any case we're going to hear from pam and nick. what is that tom hank's movie? how you do that thing you do? anyway i'm not sure who's speaking first but -- pam is speaking first for 15 minutes and then nick will speak for 15 minutes. so, without further adieu, let me introduce her. >> thank you all for this awful weather. i didn't have a joke to start with, but len put me in mind of an oldy but goody. what would happen if the economists andwet specialists switched roles? nothing. [ laughter ] >> i appear here with a lot of the question thasz are going on

here today. what nick and i are going to do, and this goes to david east introduction, we're not ready to talk about the same things that we usually talk bt a. i'm going to be providing some history to remind people of where we are in terms of modelling and modelling research. we're going to talk about the two rules of differences between them. with respect to the discussion that just happened we will happen at the end when nick talks about it. just a reminder a model is

a very simplified view of the world. the modeler has to make choices about what aspects of the economy they have to make sure to model carefully and what aspects they can simply away from because no model can solve if you're going to try to include everything about the economy. you can find that pamphlet on our web site. i think it sets the table for all the discussion that came later. if you want to get really in the weeds, read the pamphlet. what that symposium did was invited phone line groups of monitors to extent posz analyze the same sets of proproposals

assuming the same things about the kurpt law economy. we had three overlapping generation models three infill nitly live agent models and three macroeconomy metric models. and the policies that we asked them to analyze, this was driven by what was the interest in tax reform at the time. and then we had variations on them with various transition. these are just things from the pam flelt.

so, in the short run, in the vat, it was predicted that gdp would decrease by 4.2% all the way up to increasing by 6.4 prnt. a lot of the models that are here to run the longer analysis some of the ones that could produce long-rum kind of settled down to a narrower range of 1.7 to 7.5 pvrnt. so we've come kind of a long way. the thing about the other side on the program fer materials, stlfs lez e less variation in parameters than there was models in the gdp.

the choice of parameters and how sensitive you assume other matters are to the tax matters. this is one of the surprises. characterization of present law matters. we've discovered they met for three or four meetings before their final results. their results were more broadly skewed in the early meetings than they were in the end. and part of that is because everybody not even having the same understanding what lies. and the details of the proposal matters. some models were very surprised

when they put transition models in, the results applied. after that initial bit of learning, jct went out and constructed a couple models to work with. we started working with them and presented analysis that we did with them to a lot of different groups. however, e had to take in count several practical considerations. >> the real state of the art is very fancy equilibrium models. and the fancier they are particularly, they have some kind of chance element. the models that are like -- the

technical wizards have out there today. it would take two weeks to run through one simulation. obviously, if you get weird results, then you have to start over for another two weeks before you even get your first result. we can't quiet be at the cutting edge because we don't have that kind of time. >> given where we are, we have to make sure we have the tax sec tofr karktized correctly. i emphasized that again. models should have as much tax details as possible.

it's required that we provide a macroeconomic analysis of the effects of the proposal on gdp, labor, capital and revenues, basically. any bills that are reported by the ways and means committee to the house floor. and so we've done that. as it turns out, they are very small. they're so small that showing gdp affects within reasonable rounding, you get zero. for those bills, we have a statement that says results are too small to report. there are other proposals that we suspect would have been considerable.

in those cases, we quite a qualitative analysis. we don't -- since our models, we don't think -- we don't think the academic literature has enough ree research to tell us quantities. and then finally, we have full scale bills. currently, the models that we use are macroeconomic equilibrium growth model which we refer to as m.e.g. an overlapping model, we refer to as olg. you can see links to these

models on our web site. the models that we use to measure representatives were the m.e.g. model and the of the l.g. model. i'll tell you a little bit more about them. they both have bassic neoclassical from the mainstream. consumption follows a live cycle pattern. labor supply response to marginal and average changes and after-tax wages saving and con sufrgs responding to after tax return savings. that part depends on the availability of savings. both models do have cross border capital flows so that net exports affect the domestic

economy. they are our exchange rate equations in them. so let's talk about the difference in the two models. in the m.e.g. one we do have demand to adjust to supply. but in the short run that turned out to be unemployment. we've had to analyze bills that were short-run, demand stimlusz bills. our behavioral ranges are * churl. we divide our labor supply into four categories. high and low primary and high and low secondary. they tend to have different amounts of responsiveness to tax changes. and, often, different proposals affect them differently. one of the things we discovered

is when you separate that all out, you can get a very different answer. the other key thing about the m.e.g. model is the people are maya, meaning they know what the economy looks like today when they're making their decisions. they don't know what it's going to look like ten years from now. now, what this does is it enables us to model policy changes that have a growing deficit. ska one thing is e it can tell us is where the economy blows up. and that's a piece of information, as well. our o.i.g. model is kind of coming directly from what's going on in academic departments. it's con instructed on macroeconomic foundations.

it model instead of income groups, age cohorts. and the people in the model have perfect foresight. so they can look today and see a huge deficit. this is what you're always hearing about. that's why there's a lot of discussion about needing some kind of fiscal closing asuchgs. so recently we've also added a specific multinational corporation sector. on going model development, we work all of the time to do the best we can to affect changes. right now, we're double check ing some of the parameters.

we're still building our own in-house. so the rest of the presentation is going to tell you why the development of your macromodel is only half a story. getting both present law tax and the tax policy right is a lot more complicated than anyone who hasn't tried to do it in the detail that we have to do it at jct for our conventional estimates cannens e understand. and we're going to use the camp reform act. i'm showing you here a couple pages from my revenue table. it goes on for 15 pages. every one of those items we have to decide how to add

together for this model. that is what nick is going to talk about. [ applause ] >> i thought we were already on the right page. >> okay. so many people talk about dynamic analysis as though it's something that's impossible to do. note, i'm talking about die natchic analysis, not dynamic scoring. so dynamic analysis is what we've been doing for a decade. and you can argue about whether you like the results that you have or not, but we think that what we've been doing is fairly reasonable. although we don't think we're perfect, we welcome comments and discussion. many others talk about dynamic analysis as though it's a mansion kal thing. you just press a button. well, it doesn't work that way.

but one speculative guess as to why people think it does is because arthur t. clark in kwt space odyssey" is indistinguishable for magic. well, what we're hoping is that after a few minutes of looking at -- inside the hat, you'll realize there is no magic button. it's hard work. and that's what we do. pam has talked a lit m bit about initializing models to choosing models and parameter values consistent with economic literature, et cetera. we had a paper almost a decade ago where we looked at what is the impact of putting in real really schism tax economies or just an average and just a marginal or breaking it into to components that address different also pelgts of income.

and we found it's really important to get it right for multiple sources of income. in particular, for individuals, we compute average and marginal tax convert for wage and salaries. in total and for the labor supply groups that pam mentioned, high and low, primary and secondary. we compute average and marginal for interest dividends capital gains, individual returns so that's schedule c, e and f and then other. for corporations, we compute average and marginal tax rates. if you ask our corporate estimator what's the average rate or the marginal rate you spend two or three hours in a discussion about what that is. then we combine the individual rates and corporate rates to get a weighted average business rate that gets fed into the macro models. finally, both the main models

that we work with handle depreciation separately. we spoke with the estimator about liability effects and what's the capital consumption consistent with the way that the models set up. pam showed you a little bit of the 15 pages of the camp table. this is just picking one provisional most at random. it's the domestic production reduction and the columns are itm -- that's if this provision is on the individual tax model, then you'd have an indication there. individual domestic production, that is reported on individual returns. similarly it has a corporate

effect. that one is sort of obvious. for each provision, there are four macro estimators and 15 or so conventional estimators. so we can't know the details of every possible provision. so what i or somebody else ends up doing is walking around the floor and talking with the estimators about any provisions that are significant enough in terms of their score that you really want to find out, how does this provision work and what is it doing. for a lot of provisions, it's obvious whether it has just average effects or a marginal effect. for other provisions, like first in and first out inventory it gets more complicated. that has large average effects but also marginal effects and you can sort of take a look at this light and think about it more later. so at this stage we have a good idea about the details of the

proposal ideally and then the question is, can the existing models handle those details or do we need to figure out how we can modify the models so that it will handle it correctly? so for instance, the first time that we modeled the home mortgage interest deduction, we had to make sure we were modeling that correctly. some provisions you might just decide you can't model it in any reasonable way. and then you're sort of stuck. but you can't have -- you can't make models that can handle every possible strange thing that people come up with. we need to compute the average source of income and so we have atmr calculator, about 3,000 lines of code.

that modifies the existing roughly 52,000 lines that represent the individual model, which is a model that you can put in a proposal and find out how that changes liability. for marginal tax rates by income source you have roughly 48 iterations for each year because you've got nine sources of income, you need to figure out the marginal rate in the present law and you need to figure it out in proposed law. that's a couple iterations. you have to make sure that you're not leaving accidentally effects whether you are affecting income, you have to i am bed it in the data and back that out again. it's important for proposals that include base broadening that the marginal rates are calculated with respect to a broad base rather than something

that's narrow so we can represent both present law and the proposal. seemingly simple changes can be unexpectedly difficult to debug. so, in particular something that was current law where capital gains did a separate rate and the proposal to exclude what is remaining of an ordinary rate. and it takes a while to get that right. given having gotten all of this right, some people would argue -- jane will bring up a point -- we have checked to see quantitatively is jane's critique really significant or something that for the purpose of trying to get things done we can ignore as a rounding error.

and at least the sort of preliminary experiments and maybe we haven't fully grasped what you're seeing what we looked at is suggested to be down to a rounding error. so this is the -- that's -- now we've got the individual tax model atr and mtr effect. we have to look at the rest of provisions. so i think we've taken care of maybe one or one and a half pages of the 15 with provisions on the itm. the rest are not on the itm and you have to figure out the average marginal rate effect. assuming na that you've got all of that, the next step is to pick a macro model, compute the economic baseline, read in a proposed law change and then many people sort of don't think

about and/or are not aware of its importance is checking to see, does the liability change that you're getting at this state, before -- we're not even talking about running the macro effects of the policy, just calculating something that roughly correspondents so a conventional start and then checking to see if the models produce the same conventional score that the 15 pages of table show. and if it doesn't, then you've got to do some debugging. typically that means it ter rating back and forth until you've got the conventional estimate matching. now if you're convinced that the conventional estimate has been correctly computed, you start working on running alternate macro runs.

and the first question that we ask ourselves when we look at those is, does the macro effect on the revenue estimate correspondent roughly with the macro effect on the aggregates like gdp, consumption, labor supply. does it make sense in that context? and then we look at the changes in the macro aggregates for the different models and we think about whether those are behaving in a way that's consistent with what we and most people understand about how those models work. ultimately, sometimes we find there are aspects that we're -- we don't understand and so we'll go back and perform debugging modes. so often that consists of doing stacking series. so you look at the effects of the individual tax model marginal rate changes just the effect of the off-model pieces,

just the effect of depreciation and start putting them together and you're trying to understand, have you modeled this correctly or is there a step at which you've got some error in your inputs? and then typically we do sensitivity runs. so we look at the effects of different monetary policy assumptions for models that can handle that. we look at the effects of labor supply elasticities, different propensity to consume et cetera. meanwhile, while one person has implemented this on one of the models, somebody else is implementing the other models and you're going to consolidate results into a spreadsheet. that doesn't mean you're going to add together the results or take the weighted average, you're just looking at the results all in one place so you can compare them between the models and you're trying to figure out whether what you're looking at fits in the context of what you know about these

models. meanwhile, by this time someone has typically written the shelf shell of a report so they've talked about it in detail, talked about the effect on tax rates. sometimes you find an error and you have to go back and fix them but as the macro results become available, you put that into the report and that's another stage at which people can think deeply about whether these results are consistent with the models and whether they were consistent with the proposal. so we have lots of these reports already posted on the website and this just sort of lists the most recent five or six. moving forward, it's the new house rule. the new house rule as discussed before, has a trigger for when

you have to score something. but also, something that wasn't pointed out that there's a need for it but mentioned that it's already in the rule, the rule requires a qualitative analysis for the 20 years after the budgeting window. so if you have a proposal that is running -- you know causing huge deficits then what happens outside the budget window is going to obviously not be the same as something that's a revenue neutral proposal that's not only revenue neutral inside the horizon but after and what we are going to say in the qualitative analysis is still not certain. so moving forward you know we've done dynamic analysis for a long time. dynamic scoring is not something that we've done so far so that's going to be a new challenge. i think there was one other thing i was going to say but i can't remember what it is.

i'm going to finish up there. i think we're planning at this stage to move right on to the five minutes from people on the panel and that's jane gravelle and then questions can come along afterwards. [ applause ] >> all right. thanks. i wanted to mention, again our thanks to jtc staff for coming here and presenting the details of the models. i have many reactions, one of which is thank goodness i don't have to do this. but we will talk about all of that and more. we have jane gravelle alan vird from aei and center and policy priorities. each of them will speak for five minutes and then we'll all come up here and have a joint talk.

thanks. >> thank you very much. i'm jane gravelle and i want to say the views that i present here are not the views of the national research service. there may be some similarities to papers that i've written. so -- and i also want to say to our jct folks here i have great admiration for the work that you do. i know you are greatly dedicated public service and i hope if you take any questions i would raise about your analysis and the context of helping to prove it. so, this is really -- i think there are a lot of lessons we can learn from the dynamic

analysis of camp proposal which has a lot of moving parts. first, there's a big difference between individual rate cuts across the board and individual tax reform, you know, the effects are very different. so for a tax cut a really important issue is the short-run stimulus effect in the mag model and the jct model. there are a lot of reasons i think, for excluding this effect including the offsetting actions of the fed and also the lack of dynamic scoring for appropriations. for revenue neutral tax reforms stimulus effects are probably less important but it is important to incorporate base changes that are marginal in nature. and this is the question nick referred to earlier. i don't think that was done in some circumstances for example the state taxes and it's important to find out if that is important.

so i want to talk about the reasons for the differences and the results and the camp simulation. so the top three lines of this graph show the meg -- that's their in-house metric model with high and low labor substitutions and the llg. this is without stimulus effects. and you see these enormous effects between the two models and you can see the stimulus is bigger than the supply side effects on the top. when did that happen? i think there are two major reasons. one is the labor supply effect and the olg model is bigger than in the meg model. and it can't -- a little bit of it is because the embedded deep parameters that i can mention you can turn into -- this is only about 20% higher while the labor supply change is 160% higher. there's a little bit of

difference possibly from changes in capital that can't be important. so i'm not really sure what happened there. what i suspect is that the fiscal adjustments that need to be made in the olg model to make it soft have probably washed out income effects from the labor -- from the labor -- that's my guess but i don't really know for sure. i just know they are different. okay. the other reason is intangibles. you can do a back of the envelope calculation of their results and see that the labor and capital -- changes in capital and labor don't account for the increase in output. about .6% or about half of the difference between the two models seems to be due to the shift in the intellectual property and treating it as an input of the production function. sort of like fiscal capital. this has been done in a model by a couple of european economists

so it's not completely new but it is kind of novel. and i think this is not the appropriate -- an appropriate thing to include. it's not located physically. once it exists, it can be used everywhere. when a firm discovers lipitor, that can lead to production everywhere. when the patent moves to the united states instead of abroad doesn't change output because that effect is already there in output. so i think that probably should be reconsidered in this modeling. so i just want to sum up quickly what i think the main points of what i'm saying here and the things i suggest that jct think about in the future. first, there's a need for more transparency. now, jct already provides a lot of information but not enough in this case for me to understand and i've studied these models for a long time why the results are the way they are. so i just have to speculate and

it's better not to have to do that and i have to answer a lot of questions for my clients on this. so second, for base broadening, it's important to account for changes in the shares of income that are tax and measuring marginal effective tax rates. we believe those were significant and we have a crs report that looked at itemized deductions. should stimulus be included? i think there's a very strong case for excluding them. and finally, the olg model is problematic not only because of elements here but in general depicts individual workers with perfect foresight and can't measure or economy. the question is, should it been itted be used?

thank you. [ applause ] thanks. first of all, i want to thank brookings institute for organizing this conference. i want to thank nick and pam for the exposition on what jct has done and is doing which i found extremely insightful. i only have five minutes so i'm going to mainly comment about two features of the estimate of the camp bill. each of them probably has implications for evaluating the camp bill but also have implications for thinking about what dynamic analysis and dynamic scoring can and should do and, of course, that will be my focus. the first feature of the jct estimate of the camp bill is that business tax reforms made in that bill do not increase the capital sought or do so to a very slight extent. there's a variation in estimates but they are negative that the bill would have reduced the capital side. i think that's informative from

the established the standpoint from economic policy. the initiative is relatively clear and it pertains to the nature of rate reduction offset by base broadening in the business tax context that the rate reduction benefits existing capital as well as new capital while the base broadening almost exclusively falls on new capital and so that mixture tends to put an -- increases the burdens on new investment and gives a windfall to the existing capital. we've already got it so there's no incentive effects on that score and therefore we would expect that there probably is a reduction in the capital and the estimates do show that. marty sullivan has written about this feature of the estimates. now, i think it has some policy implications for whether you want to do that type of reform although there's many other factors to consider. what i want to emphasize here for a moment is the implications for what we're going to get as we start to do dynamic scoring

which is we're going to see the different kind of tax measures, different kind of tax cuts and tax increases actually do have different effects and i think that's one of the important contributions that an analysis can make. it's not to make all tax cuts look good or all tax increases look bad but to sort out which types of tax cuts have the biggest effects on growth. so things like taxes that apply to old capital and taxes that apply to new investment distinguishing that is very important. so another thing that drew my attention also pertains to the effects of rate reduction offset by the labor supply. jane has already talked about this some. some of the estimates that jct found show what i view as a very large labor supply effect. 1.3 to 1.5% increase in the olg model. of course, as we know jct did

provide quite a lot of information about this estimate but i think i would have to echo jane's comment that it's not quite as much as is ultimately needed. the explanation that jct put out is that it reduces marginal tax rates on labor. as a first cut, i wouldn't expect that to be the case because it's divisional rate cut broadening combination and with standard economic assumptions you expect that to lead the labor roughly unchanged. for example, if you have an economy with 40% statutory tax rate, people spent half their wages on apples and half on oranges and apples were exempt and oranges were taxed you can lower the rate to 20% and tax both apples and oranges. of course, the effective tradeoff at the margin between leisure and consumption would still be 20% so you would expect

a zero labor supply effect. many of the reasons why it would not be zero. there's what economists called on and it's not necessarily an indication that anything is wrong. but a fact this large cries out for an explanation and so i guess the implication that i draw here for dynamic scoring and analysis is the need for greater trans parnparency, to breakdown what was the change in the tax rate and that has to be done for different tax households and then showing how the base broadening marginal effects, how much of that was taken back from that channel. so those are the two main features of the camp estimate. let me just close with a few seconds, mentioning something that nick actually touched on. i do think that section 8c1 of the new rule is important where it calls for analysis going out an extra 20 years.

i think it's a mistake if we start obsessing too much about trying to nail down these macroeconomic effects within the ten-year window and ignore what is really important about the long run and i know that the long run is harder to estimate and that's why the rule says go out only 20 years beyond the ten and why it says to do it in qualitative terms instead of quantitative terms but i think that's an overlooked part of the rule that may be ultimately as important as the rest of the rule. thank you. [ applause ] >> hi, how do we avoid misinterpreting them and if a dynamic score of a policy looks different than a traditional

score in some way, the temptation is going to be simply that means the policy must be good for the economy. but the heroic assumptions and large gaps in the model i think means that we need to treat the dynamic scores with much more caution than that when we interpret them. so if you would go to the media releases, the tax plan had a real growth impact that would have led to 700 billion in extra revenues over a decade. but, of course, that was just the high end of jct's range of estimates and the low range was a much more modest 50 billion. it came from one run of the olg model that's been mentioned. and to get that 700 billion jct had to pretend that there would be large tax increases and transfer cuts baked into the baseline. so those are things that are in addition to the camp plan itself.

and it's because, as you've heard, olg simply doesn't work if you don't assume that lawmakers enact additional deficit reduction to stabilize it as a share of gdp over the long run. so that much touted 700 billion number did not talk about the growth effects of the tax plan and may even predict. so when we get a single dynamic score that's going to be incredibly important to understand whether and how the olg model for example contributes to that score and the predictions about future congressional law making that are driving that result. now, the house rule requires a single point estimate and to make any sense of that score at all, it's going to be really critical to know that outputs from the different models and si shum assumptions that produced that single score. and echoing what some of my other panelists have said it's

going to be crucial for jct to show results from a range of models and assumptions so we can understand how that score works with the other models. for example, if olg does end up being used, it's going to be important to understand what would happen if you made different assumptions about how future lawmakers deal with deficits and when they deal with deficits. another thing that is highlighted, we need to know not just how tax reform affects the economy and the budget but how it affects people at different parts of the income distribution. jct's distribution analysis of the plan showed that effective tax rates faced by people at different parts of the income scale was before and after the plan. and camp relied on those tables to claim that the plan was

distributionally neutral. but unlike the 700 billion of revenues from the growth estimates that he cited, those distributional tables didn't bake in deficit reduction from future congresses. so if you had put that deficit reduction into the distribution tables, they would have looked very different. so when we get a dine flat mcscore and a distributional estimates, it's going to be important whether they show the post reform set of policies. i think a final potential pitfall in interpreting these scores is, you know, the current estimates, the current models have some pretty big gaps and jct can do some things off model to try and deal with them but, for example, there's no explicit modeling of human capital. if you had a plan that really boosted skills and training, the high productivity from increased human capital accumulation

wouldn't necessarily show up in the score from the base models. so that's despite the evidence that doing such investment is associated with productivity growth and increased investment. likewise, there's a bunch of sectors not modeled. we might want to talk about that a little bit more on the panel. and jct is doing the very best with the very best models that they have available but we have to watch out for that particularly if a reform affects a particular sector. so while lawmakers -- some lawmakers will simply want to treat a dynamic score as proof that something is good or bad for the economy and the people in it, i think we have to keep in mind the uncertainty, the flaws, the heroic assumptions that go into these models and those things may mean that that conclusion is simply not sound.

[ applause ] >> okay. while we're getting set up, let me thank the speakers again and to say that you all -- frequently the moderator starts the discussion among the panelists but you all have been very patient and i know there are a lot of questions out there so we're going to turn to exercise my moderator prerogatives. so questions. >> and i, too, want to commend

brookings institute for having this topic and thank jct for coming out and being so open about what's inside the black box, giving us a view under the hood. i want to reiterate a point that jane brought up because perhaps not as much as jane, i have been studying these results and i cannot figure out what is going on so i'm going to take advantage of this opportunity to ask jct about the multinational sector. if i have it correct, in their analysis of the camp proposal intellectual property migrates because we lower the tax on the united states on intellectual property and raised the tax on the foreign side, intellectual property migrates back to the united states. and as i understand it in the model, this increases productivity and economic growth in the united states. is that right pam? >> pam or nick i just want to -- just pin the mike.

just put this in the context of the issues we've been discussing in elasticity that we really don't have great evidence on but that nonetheless is critical to the growth effect. >> let me put it in terms that everyone can understand. if you have a multinational which has a domestic factory and a foreign factory, because of the tax changes in the camp plan, for example, intellectual property is know-how or expertise, that expertise is going to move from the foreign location and reduce output there and then come into the united states and increase output here just because of a legal recharacterization of where this property is located and i find this totally unrealistic but i could be missing something so i just want to ask for

clarification. >> so, i think one thing that could have been improved in our report was describing what we meant by i.p. a little better because it's not just intellectual property it's really intangible property. and the multinational corporation sector, with that shifting, based on the paper that was cited earlier and, yes it does have that effect. we are as we mentioned always trying to look for whether or not models need to be altered and we are researching right now what we think should be done about that. >> yes? >> just to keep the tax analyst going, i'm a capital reporter. this is also for nick and pam. and i'm not sure if this question is too vague, too complicated or something you already covered, but how do you think about the issue of

crowding out because objectviously the way it's financed can affect the economic impact and i've found, so far being in the tax world, that's a big factor that people kind of dance around or don't fully acknowledge when they are talking about how they view the economic effects of tax cuts. >> okay. so particularly in the meg model where we don't have to make any kind of closing assumption whatever crowding out is happening because of the proposal is going to have the normal sort of effects that crowding out has. it drives up interest rates and is therefore going to depress capital formation as you move down the horizon. for olg we try to make the -- there are two things. first of all, the closing assumption, because we are trying to model tax policy, we try not to use a counterfactual change to tax policy to make the closing assumption.

second of all so what we have been doing is looking at transfers but another possibility is to change government spending and so we'll -- you know, that's something that can be reviewed how exactly to split that up. but the second thing that we do is the closing assumptions, we try to make those happen after the end of the budget horizon. now, that means the crowding out can happen inside the horizon and have its normal crowding out effects but because olg is forward looking people do anticipate, for instance either that there is a change in transfers after the end of the horizon or a change in spending, which ever one you're using and that can affect that because of the anticipation. >> this is a question to the panelists or the two of you from jct. kind of a bottom line question which is so given what we've learned over the last 30

minutes, which is very helpful and as i've already said, very under the hood with most of us do you think that the accuracy of the budget process is legitimately enhanced improved by a rule that forces estimators -- and i would argue nudged by partisans -- to make one choice, to choose -- in other words, dynamic analysis, yeah, that's different than dynamic scoring and i guess i'd like to hear your views on whether this -- whether the fact that we're talking about choosing one score given what we've just heard, would actually improve the accuracy of our process. >> let's have our panel address that. >> want me to start in okay. i am actually on the record as saying that i don't think we're ready for dynamic scoring when i testified before the budget committee because of the variation that we've seen here and all of the moving parts.

i think it's certainly possible that the camp proposal had a zero effect on growth, or possibly a negative effect depending on how these sort of things that i find questionable. if you've got an estimate for one -- one estimate is 16 times the other or 15 times the other i mean, i don't know what kind of answer you have there. and i think when you go back to -- for many things, going back to what doug was saying, i think for many things, the static scoring is very clear. if you have rate cuts, you have a clear set of data for doing all of these things. but until we can get some kind of a consensus about macro effects, you know we're kind of forcing these guys to do or pam and nick is a guy, sorry, if i was still in the south i'd say y'all but i've tried to get over

saying that. that you are kind of forcing them to do something that is almost impossible to do. so that's how i see it. >> so it is an excellent question. i think it's a judgment call because we've certainly seen how difficult and how uncertain this process is. i think it's certainly increased our respect for the people who do this -- who have been doing the dynamic analysis. i realize if we go with the dynamic scoring, it's important to keep in mind how modest this real change is and i think in this stage of the process, after all of the years of doing dynamic analysis it should be useful to be doing one or two dynamic scores a year for the next congress, the next two years and see how that process plays out and improve that. we ultimately do want to get to a point where we are taking these effects into account and this seems like a very modest step towards doing that. maybe what's worth stressing is that there just really is a con

sen shus here census here on a lot of things. we all agree it's difficult to do it right. but we also agree that the overwhelming majority of bills that we should not yet try to do this for and so the only issue of dispute is, should we be trying to do this for one or two a year? and although i think it's possible for people of goodwill to certainly disagree, let's give it a try for one or two big bills a year. >> so the problem, i think, with that response is that the house rule allows the chairman of the committees to essentially designate any bill to which a dynamic score must be produced. so the 2.5% rule that we've been talking about can be blown away by the chair of the budget committees. alternately, if they think it might not give a favorable result, just split up the bills. not report it out of committee. so i think -- i agree with jane on the substance and i think i

disagree on the fact that the process is going to protect jct in the sense that there's only going to be two or three bills a year. it's essentially at the discretion of the committees. >> i want to add in that it's also important to note that when we do a quote static score unquote, we're really doing a stat tech analysis. we're averaging the effects of a bunch of static scores. >> so without opining at all on the substance of what everyone has just said, just a little sort of clarifying comment is that we anticipate that for something like camp you'd still have exactly the same 15 pages that were already published but now there will be an extra line at the bottom that just has -- >> while you have the mike, can you clarify -- >> i'm not speaking on behalf of the committee. this is just my comment. >> can you comment on whether

it's gross or net? >> we understand it should be gross. >> the net? >> the text of the bill says gross. the bill does say gross. >> we know what it says but we don't know what it means. adele? >> i am adele morris the policy director for climate and energy project here at brookings. my question is about the scoring of a bill that's got a lot of complicated elements in it that are not simply just fiscal policy or tax policy. so for example, let's suppose we had a bill that imposed an excise tax on carbon content and fossil fuels for example, and some of that revenue buys down business tax rates, for example, so you've got the kind of question of how you're going to dynamically score a tax swap maybe with additional elements in there. but then what if also, there's

a regulatory reform in there that says we're going to suspend clean air act regulations on stationary sources? how do you or do you deal with that and also, in that context would there be a change in how cbo would score an excise tax because they have the standard 25% haircut on the gross revenues of excise taxes? would the dynamic score change that? >> all right. let's get an answer to that. >> i would hate to be confronted with that issue but excise taxes do have allocation effects and labor. an excise tax is like a labor tax. so i would treat it as a labor tax and also changing the allocation -- i would be clueless. i would be sorry for whoever would have to do that but i'm not sure they'd have to do that. >> thast's essentially changing the baseline.

eric? >> thank you. i'm eric. i had a question about how these results get presented. i just heard nick say that to close the model on the camp proposal they cut transfer payments. so what we had in the dynamic score was not an estimate of the kampb camp proposal, we had the camp proposal proposal plus a cut of social security benefits and that would play out very differently in the public mind if that were the way that it was presented. i'm not questioning the accuracy in my comment. i guess my question is, should people -- since models require some closing assumption, should members, in order to get a dynamic score, be required to specify the closing assumption they would like used? >> it's simple yes or no question. thank you. >> can i correct the record for a second? so there's a -- that's a kind of common misconception on the camp

proposal in particular. it is true, we have to have a standard closing assumption for the so-called present law baseline in order to get a present law baseline from camp. and what is in there -- and i will concede we probably need to re-examine how important we think that might have been to the estimate -- but what is in there is some of -- all of the above approach where the economy present law economy is kept on track where debt does not grow faster than gdp by a combination of increased taxes and reduced transfer payments. now, that is separate from how did we analyze the proposed camp reform. so one thing you want to remember about the proposed camp reform, it was explicitly designed to be budget neutral. that after all of the base broadening and tax rate changes were accounted for, there would

be zero effect on the deficit. in fact if you look at our 15-page conventional revenue table, that's really close. >> all right. so -- >> let me finish. >> one can imagine a different proposal that's not revenue neutral. >> can i just finish this part? >> yes. >> so -- but if it's under conventional analysis, deficit neutral and something that causes growth, then you might have the reverse situation where you would have surpluses growing faster than gdp which is also not fiscally stable and so you would have to make an assumption that pulled down that. now, for the camp proposal -- and this is all in our analysis this tells me people are looking at the tables and not reading the verbiage because it explained what the baseline was and also what the closing assumption was in the proposal.

we had to assume a slight increase in transfer payments in the future to make up for the growth that was being generated. >> all right. thank you. this question is not about the camp proposal. i think the question is if someone proposes something that's not revenue neutral or budget neutral, should they have to propose a closing mechanism, a way to raise the revenue? that's how i interpret eric's question. >> or give up the olg model. because if you do a tax cut and the correction with transfers, that's going to be a very different outcome or correction for government spending and sort of an even bigger effect of the tax cut if you assume taxes -- the labor supply response. i think they chose a fairly benign change although if it had been me i would have changed government spending. >> let's make this a general

answer and not a camp -- >> i think if the member specifies it, it's actually written into the bill of course, that's what should be done. i think if it's not in the bill as doug said earlier there should be a standard assumption. and that should apply to everything and i think it also -- that should be incorporated as well. >> the problem with that though, you're then assuming things that policy makers will do things that they have not said that they will do. >> but they'll have to do something. >> to avoid that assumption. >> but you can't just avoid it because budget requires they haven't done something that they said they are going to do. they haven't said that they are going to fix the fiscal situation but they are going to have to. >> but then the congressman says my proposal would raise growth and someone says that's because you're cutting food stamps and the congress says i never proposed food stamp cuts. i've just proposed tax cuts. so it allows congressmen or

senators to talk out of both sides of their mouth. >> i think you need to make very prominent what the standard assumption involves and that should be part of the analysis and if the member says that's not what i want, then the response should be, why didn't you put in the bill what it is you do want. >> i want to highlight, the standard of assumption can flip the sign of growth effects. if you'll if you finance with future welfare spending cuts you'll get maybe a positive impact. >> just one more thing about this and then i won't say anything else, is that i think the most benign assumption for a tax change is a change in government spending. because at least that allows the income and substitution effects of labor to play out in full. so you could just say let's try to choose what's going to least disturb our analysis directly. >> i think going back to what i was saying earlier whatever goes into the single estimate,

however you choose it, if you make the congress say whatever it is going to be, or you leave it up to jct, i'd love to see the sensitivity analysis what would happen. >> okay. one last question in the back. >> hi my name is ricky. i'm a student at the university of georgia. it's nice to meet dr. gravelle up there. it seems to be all or nothing. would it make the panel more comfortable if we kept dynamic scoring and use it and see where the margin of error or the difference is since dynamic scores is going to provide a larger estimate? >> very short answers. >> yes, but i think pam said that's their plan. >> the conventional score should

be disclosed. i assume it will be. again, it's not all or nothing because we're talking about on a small number of bills. it is my hope that the budget chairman not -- use the discretion that the rule mistakingly gives them. >> all right. i want to thank all of the panelists for coming. [ applause ] >> so i think we've established here that i'm not sure about the incentive for academics to make public policy better but i think we are establishing the case and i'm serious about this, that if congress is going to proceed with this resources at jct to explain what they are doing are needed. and i think that communications

challenge of this thing is evident from the panel this morning in addition to the idea that giving members of congress incentive to do things that would increase the rate of growth doesn't seem like the worst idea we've ever come up with. but one of the very interesting parts of the house rule is that it also applies to certain spending bills. that is mandatory spending or entitlements that would increase or decrease spending by about $40 billion a year so it would have applied, for instance to the affordable care act, it would have applied to the presumably to the fiscal stimulus bill. it does not apply, however to ordinary appropriations and there's some debate about that. so we wanted to avoid focusing only on taxes because spending matters, too, and we have three people here to help us do that.

first is donald meagher oa rchl ron. and don did a stint at both the cea and was acting director of cbo in 2006. steve mcmillan is now withsteve mcmillin was omb of the bush administration and jared bernstein, senior fellow there and did a stint of chief economist in joe biden's office. i want to start with you, donald. when you were the acting director of cbo, you were confronted with an immigration bill and you were involved in trying to figure out what the macroeconomic effects would be and how to dynamically analyze it. i don't know if you dynamically scored it. for people not into the lingo that means you tell them what the macroeconomic effects are.

dynamically scoring means you actually figure that into the official price tag. tell us a little bit about the immigration decision you had to make. >> first, i want to emphasize that i think the dynamic scoring debate has so much focus on tax that the spending and regulatory policy angle of it has not gotten enough attention and it's where these issues are going to be very important. in 2006, much like happened in 2013, congress considered a major immigration bill that would change the size of the u.s. population, changing the size of the u.s. labor force by several million people. at the time, the joint decision of jct and cbo was that you couldn't ignore that in doing the score. if you tried to follow the convention of holding macroeconomicsing a gre macroeconomics aggregate, 6 million new people join the labor source and assume they are

all unemployed, that would literally be insane. so for large immigration bills the conventional exception from the convention is to in fact take account how those immigrants would affect the macroeconomy but try to do it in a way that's respectful to the distinction between kind of conventional scoring and macro dynamic. there's three categories of scoring that happens. some folks have used the phrase static which i avoid. they recognize that people respond and so if you change taxes, spending, people will respond to that. if you pay doctors more, they might work more or less. there's a debate about that. if you tax cigarettes, right people will smoke less. for major immigration bills, what's taken account of are the first order direct effects that happen to the macroeconomy. so you have a bigger labor force, you have more people in the workforce, more wages

overall, that's going to result in more tax revenue and more people around and so you have more spending on various spending programs and you're going to want to track through the net of that. but what the cbo and jct scores of immigration do today, they don't then take the next step and track through every indirect macroeconomic effect that would follow. and so for example, they only account for some of the increased increased investment in immigration and they analyze those in a separate report. you have kind of a score that includes some change in the macroeconomic situation and then advisory report, much as we now have for tax bills that tracks through the additional things that would be incorporated if you wanted to go fully macro dynamic. >> and under the new rule works that change? >> you would try to incorporate all of them assuming it triggered the .25% of gdp.

>> jared, much is often made of the inadequacy of the american investment in infrastructure public investment and in infrastructure. as i understand the rule it would be hard because they exempt appropriation bills, but in general do you think if we're going to score tax cuts that we're going to score infrastructure and wouldn't that make the case even more attractive? >> well on the surface it sounds like it would but in practice i fear that it wouldn't and i fear that it is even harder to do that sort of thing than the tax kinds of scores that we've been discussing so far. appropriations -- this has a lot to do with budget process and the way that cbo scores such measures but appropriations are too uncertain, generally, for

scores beyond a year. that is the cbo can't sit here and say, here are the appropriations that we see in year ten. they can't even say the appropriations that we're going to see in year two. if the dynamic score showed positive effects for appropriations it wouldn't show up beyond cbo's estimate of when those appropriations would be in place. a good example is the highway trust fund. how could you possibly if you're a cbo, figure out what is going to happen with the highway trust fund when it's going to allegedly go bust in may just a few months from now? we can all assume a patch and probably a patch will be in place but i think it's a lot to ask cbo to do that. >> if i say i want to raise the gas tax, i propose a bill to raise the gas tax to 50 cents a gallon and use all of that money for infrastructure couldn't that be dynamically scored? >> um, that could be dynamically

scored and depending on the extent to which cbo would believe that would be in future years that might work but another bias is a low productivity bias. so a cbo writes for federal investment, estimate is additional yields half of the typical return on investment completed by the private sector with an average delay of five years. so cbo assumes for this kind of investment crowding out of private investment. and so that, too, would be a bias against the spending impacts of appropriations. and, in fact, just speaking a touch more broadly, i tried to think about the positives and the negatives of dynamic scoring on the spending side and all i could come up with were negatives.

the idea that there are biases with spending scores being positive throughout the process. the most important one -- just let me get it out on the table -- is basically -- and this is from work that richard cogan did. i didn't know about this. it's a bias that would lead dynamic spending scores and the budget forces allocations on the revenue side on the revenue target and the house and senate rules disallow taxes to go below the revenue floor or appropriations to go above committee's allocations. under the rule that we're talking about the house dynamic scoring rule if we were to move towards scoring discretionary, any positive growth effects would be scored almost entirely as extra revenues just by the

rule itself, not as something that would give you room on the spending side. you would only have the ability to cut taxes more, not to increase spending based on what would pop out of the model. >> steve we know that when congress does things what we have found from listening to the jct folks on the tax side is if the new house rule had been implemented ten years ago, the quality of what you would get would be at one level but they've been working diligently all of these years to improve that quality. there's been many level of

analysis to make that happen on some types of spending programs and when that type of analysis becomes more relevant to the policy making process i think you'll see investment and improvements in that. now it may be that that in some areas as we heard on the tax side, we just can't figure it out. it's too uncertain. the answer for purposes of this exercise is zero. i think that's going to happen quite a bit. to the extent this becomes more common on spending bills. i just want to comment on one thing jared said. it's important to distinguish between the analysis itself and then what it is used for. so a good analysis on a large increase in structure spending will inform the debate and tell policy makers how they ought to vote or what the results of

their vote would be but that doesn't always flow into the rules for consideration of legislation. now it is true that the feedback if one assume there's a growth affect from infrastructure would be primarily on the revenue side but if you pass a budget resolution in the congress that assumes an increase in infrastructure investment, and you dynamically score that that flows through all of the aggregates and you'll find that the revenue aggregate takes into account those higher revenues from your investment in infrastructure so it only creates the bias in the policy to the extent that a budget resolution leads to that result. i think you're assuming that somehow dynamic scoring, this rule, is going to lead congress to increase its spending caps or throw away sequestration and i certainly don't make that assumption. >> well -- and again what to

the same set of facts lead different policy makers to conclude is the right policy to pursue? in this particular case, i think when you're looking at a republican congress or republican senate, it's not likely that that analysis is going to cause them to increase the spending caps. every six years or however often we do highway bills these days out comes a rather simple model that department of transportation put back awhile ago saying every billion dollars on highways is 17,000 jobs or 42,000 jobs or 27,000 jobs, depending on what year you did the estimate or the last time the model is updated and people use that on both sides of the aisle to advocate more infrastructure spending. sometimes that argument is successful, sometimes it's not. >> i think those are fair points. we should to more infrastructure spending and cbo and everybody else should do dynamic analysis on that. the concerns that i'm trying to

express here is that is one corner on a set of changes to this whole dynamic scoring debate that look to me like it would really lead to a bias in approving more tax cuts than i believe we can afford. >> donald, a lot of the bills that would clearly trigger the rule involve health. if it had been in place for the affordable care act, for the medicare and drug prescription bills, if we ever do some big whole sale change to medicare, put yourself in the position of the congressional budget office. how challenging is it going to be to them to come up with some judgment about what affect big changes to health policy have on the economy. >> obviously a big challenge but new area to look at. i any it's useful when you look at dynamic scoring or any major bill as jane mentioned there's some short run sim will yous affects either positive or

negative. stimulus, in the wronger run there's the supply side affects are you doing something to increase labor force increase capital. for example something on the held health side improve the quality of the american work force and something that changes the budget balance over the ten year window into the future. is it something that's crowding out or crowding in private investment. >> cbo in principle would want to go all of those channels. is there a short run affect boosting or harming the economy. is it is a reform what are the affects on investment and then try to track through how of the changes might affect labor supply. we know from the discussion of cbo's analysis of the aca for example that there's some interesting questions about how the implementation of the aca may have affected the number of people who choose to be in the labor place at any given time. >> could i make a comment about that?

i agree with done's's comments. i think the aca's labor supply example is a very good one of the kinds of biases that i worried about in this case the margin utility bias or social bias is nontrivial. so they come out with the aca score and it does have this labor supply affect part of which is by dent of allowing people to move from full time to part time work if that's what they want to do because they can now afford to get health care, sometimes subsidized in the exchanging. you're releasing job lock. i'm not making this up. this is something cbo said is part of the mix. well, in the hurly-burly of the debate where gdp is kind of elevates a

elevates above everything and labor supply, that's a big negative. it is not don's fault or cbo's fault, this is what happens when metrics and labor supply are elevated above others. the social wellfair of the nation was enhanced by unlocking job lock. so a diminished labor supply is not an obviously bad thing unless you're fetishistic with labor supply or gdp. i also think increase of voluntary part time work, unlocking of job lock is a very good thing but i think it's outside the scope of dynamic scoring in a way that's dynamic. >> but that's just saying that congress should not make all of its decisions based on the scores. that they can decide that it will cost something but it's worth it because it has goals. that's true of any of the scores. doug elmendorf said when he did

the health care evalvation for clinton that this should not be the only part of the criteria. >> i think if wise people were suggesting these considerations i would feel a lot better. >> you're saying that congress doesn't meet this criteria. >> correct. >> so this is a pressure you feel a lot at cbo so the b in cbo is budget and the primary goal is to provide scores that guide the budget process and the dine age dynamic discussion should be about making those the best possible. it is not the congressional office or the social wellfair

function office. it sometimes would be fun if it were. >> welfare. >> this is a very important challenge to work through particularly if we expand what gets scored which is the budget number has to be treated in the overall policy discussion weighted against other things. >> let's say that it's 2009. we're in a deep recession. the president proposes a major fiscal stimulus. the arrra. the chairman of the budget committees tell cbo we want to know the dynamic score of this. would that have been a good thing to do? >> absolutely. certainly we saw some numbers from the cea projecting what they thought the outcome of the implementation of the president's policies would be. there's now a record to compare that to. but in terms of the relevance of what that dynamic score would have been on the legislative process, on the enactment of those policies, the question of

whether you were deliberately increasing the deficit by $850 billion or $600 trillion was not something that was really going to affect the outcome of that particular debate. there was a deliberate policy choice that we need to expand the deficit here over the next couple of years to try to get some beneficial affect in the economy. so if the feedback affect of that was baked into the official score, then people would have to go into their talking points and, you know, scratch out number one and put a different number. >> but it would have looked like a smaller price tag and that might have paved the way to a bigger fiscal stimulus. >> perhaps unless you're up against some -- you know, arbitrary threshold like the t word as opposed to $999 billion, the