A Conversation with Dr. Lori Taylor

Dr. Lori Taylor is the new head of the public service and administration department at Texas A&M University, and an economist who studies school finance and efficiency.  And she’s played a key role in shaping public policy around these issues over the past few decades. In addition to advising legislators on aspects of our school finance formula, she carried out a study that was used by the state in its defense during a lawsuit before the Texas Supreme Court.  The study sought to determine how much funding is necessary for an “adequate” education. Arriving at an answer proved to be pretty complex, but luckily she was willing to share what she learned with Commit communications director Rob Shearer and I.

Dr. Taylor: There are a lot of moving parts in that kind of an analysis, and there are a number of approaches that folks take to answering that question. The one I use, the one I'm most comfortable with is to look at the current pattern of outcomes and resources and essentially use that estimated relationship between inputs and outputs to predict what the inputs require to achieve a given level of output. The most difficult part of that analysis in many ways is determining what level of performance is consistent with adequate, because adequate is, in many ways, a political concept. It's not an economic construct. And so if you tell me what you want to accomplish, I'll tell you what the people who are accomplishing it currently spend and whether it's reasonable to expect that other folks could spend, could achieve similar goals, spending similar kinds of money.  

Joshua Kumler: So the Texas case was really interesting because there was another survey, I almost said similar, but they were actually very dissimilar-

Dr. Taylor: And they're not- neither of them are surveys. I would call them studies.

Josh: Okay. A study. So your study came to a number that was something like $861,000. And another study that was carried out by the plaintiffs in this school finance case was $457 million.

Dr. Taylor: Right.

Josh: Your number ended up being more borne out by what then happened. So I'd love for you to talk about how specifically it was borne out and also just how two studies can come to such different results and what that tells us about what is the best going forward with things like this.

Dr. Taylor: So the studies that you're describing were conducted at slightly different timeframes, in 2004, 2005. The study that my colleagues at Texas A & M and I worked on on behalf of the legislature, was designed to describe the costs associated with achieving particular levels of student performance, such as saying, we recognize that greater student need takes greater resources, that there are economies of scale in education, that labor costs are higher in the cities than they are in the rural areas. You have to take all those things into consideration to figure out what it costs to operate a school. And so we did some analysis work and then a lawsuit was filed, and the plaintiffs brought forward two competing studies. One of which was a data-based analysis using regression statistics, basically the same technique that we had used. Another was constructed using what's called a professional judgement model, which is where you bring together people who have expertise in the area and ask them to design a model school, and kind of roll up to the cost of operating a district from your model schools. So, in some sense, there are two approaches: from the bottom up, build a model school and then I'm going to replicate it all over the place and say that's my estimate, or from the top down, these are what's being spent and what's being accomplished and if they can do it, I can too. So we're going to estimate what it costs to achieve a certain performance goal by looking at the places where that goal is already being met. So you surround the data with the two methodologies. The estimates that came from the plaintiff's experts were very different from the estimates that we generated on behalf of the state. And there are three reasons why it turned out that way. One of the important differences in the modeling had to do with the assumption that was made in projecting how much it takes to operate a school. My team said, you know, if the model says your district would need to spend  $8,000 per kid to operate a successful school and you're currently spending $8,100, then we're going to say that you're currently receiving sufficient funds to be able to achieve that goal. The plaintiff's experts basically said, if you're not currently at the performance metric, then we're going to estimate how much additional resources you would need to get to the performance metric regardless of what you're currently spending. So anybody who was below the standard of performance they were trying to cost out, automatically was going to have to have more money in the model regardless of what their current level of spending is. And that simple assumption about whether you believe that improving performance will always require additional resources, or whether you believe that if the model predicts your level of spending is such that you ought to be already at the level, then we're not going to say you don't need any more money. And that was really the big difference in between the two. They generated remarkably similar estimates of costs for a lot of the districts in the state. The other big difference between those two empirical models had to do with what's called a pupil weighting assumption, which is really arcane and deep into the weeds of the data. But basically what it means is that the plaintiff's experts assume that the relationship between inputs and outputs was going to be influenced by the number of kids in a district and their statistical assumptions that essentially you have to hit the number right for Dallas and Houston because they're huge. And our approach, which naturally we thought was better, was to say that there's information about the relationship between inputs and outputs in every district. And just because Houston is 200 times bigger, doesn't mean there's 200 times more information in their spending patterns then there would be in the spending patterns of a smaller district. So we did our analysis at the district level and the plaintiffs did their analysis at a pupil weighted district, which is along the lines of a student level. So that was another big difference. The third big difference in estimation had to do with the way we measure student performance, we used a growth model, they used a levels model, lots and lots and lots of little differences. And so, with the benefit of hindsight, we wrote up an essay on best practices in a cost modeling and pointed out to folks that all three estimates were wrong, because the plaintiff districts, every single one of them got to the performance threshold we were all trying to cost out. And most of them got there with an actual- given inflation- real cut in spending. So our estimate, which was in the hundreds of thousands rather than the millions range, was still too high.

Josh: So do you think, and you know, obviously it's getting more into speculation, but you've also kept up with these issues. If you were to carry out a similar study right now, do you feel like the results would be fairly similar or different?

Dr. Taylor: The results of a cost function study, which was the methodology that we used, tend to say that to achieve what outcomes that we on average currently achieve, costs on average what we currently spend. And that the real issue is in the distribution of those resources, that some places are more challenged, have needier student bodies or higher labor costs, so they need more resources than current funding levels would provide. And other districts are in more favorable circumstances. They could probably accomplish those performance goals.

Josh: Talking about ways to make things more efficient, you've written about a number of different methods. One of the big ones is adjusting the cost of education index, which, it was pretty crazy to me to find one essay you had written in 2003 on that topic and another one from 2015.

Dr. Taylor: What's pretty interesting is I managed not to use exactly the same words both times.

Josh: So if you could kind of go into a) just what that is for people who may not know, but then what it is that you're hoping to see adjusted and why you think it has taken this long.

Dr. Taylor: The bottom line is understanding something that just about everybody in the business community gets, which is labor costs are different in different locations. Cost of living is higher in some urban areas and in other rural areas, there's amenities that make a particular community really attractive or a lack of amenities that make it particularly unattractive to locate in a specific community. As a result, labor costs are not the same from one part of the state to the next. And Texas is one of the states that has recognized this difference and has known about it and incorporated it into the funding model for decades. And the way that we operationalize those adjustments now as to what's called the cost of education index and the cost of education index modifies the funding levels of that districts receive to direct additional resources into high labor cost communities in the state. And all that's Hunky Dory and a very attractive feature of the funding formula. But the index, which describes where are those high labor cost communities and how much more does labor cost in these communities, hasn't been updated in more than 25 years. And so it's reflecting the pattern of teacher compensation in 1989. And things have changed a bit in Texas. One of the biggest changes in Texas has been that the housing costs have been going up statewide, but it's been really going up rapidly in the more urban areas. San Antonio, Austin, if you know anything about Austin, you know that the cost of living in Austin has been going through the roof. And as a result, there has been a change in the pattern of labor cost differences in the state. If you fail to recognize those labor costs differences in the state, what you're doing is disadvantaging these school districts and communities where a certain level of teacher salary just isn't competitive. And as a result of those districts are unable to hire the high quality teachers available to other districts at lower cost. So for pure equity grounds, you need to recognize that labor costs are higher in a Dallas or a Houston, San Antonio, and they need more resources. And the difference between Dallas and Dalhart has widened over time. So I've been advocating since I started on these projects in 2000 that the state should go ahead and update the cost of education index. I should probably fess up that I was involved in the original estimation. In my cradle. But I was involved in the original estimation of the cost of education index in '89. And, you know it was an incredibly sound idea, it's still an incredibly sound idea, it's basic supply and demand economics. But the pattern has changed and the funding formula needs to change as well.

Josh: You wrote about the difference between resource equality and spending equality-

Dr. Taylor: Right.

Josh: - which I found very interesting because we have been exploring in a lot of different facets that difference between equality versus equity and to really achieve true equity is sometimes you can't have completely equal inputs, and was pleased to learn you've written a lot about that as well.

Dr. Taylor: Yeah. No, to me it's a profoundly important truth about education, which is that equity and equality aren't the same idea here. That you need to be thinking about real resources, not dollars. And the dollar doesn't stretch as far in high labor cost communities as it does in others. You also have to recognize that there are large economies of scale. It's gonna take more per kid to operate a school district that has 300 kids, then a school district that has 30,000 kids, and it's because in much of rural Texas you can put the kids on the school bus for hours and you still can't sweep up a cost-minimizing, cost-effective classroom. And so you're going to have labor costs that are much, much higher than in a more densely populated urban area where all of your classrooms can be close to optimal size, where your labor costs are much lower. And if you fail to recognize that then you are going to under resource the small, by necessity, parts of this state.

Josh: Yeah. And related to that, another thing that is sometimes proposed in order to get at creating more efficiency and economies of scale is school district consolidation, which is something else you've written about, and I was surprised to see you advocate against.

Dr. Taylor:  Yes.

Josh: So could you talk about why you would not recommend it?

Dr. Taylor: Well there are a couple of arguments. There are clear economies of scale in education. But they tend to peter out at a school district size of about 11,000, 20,000, say, upper end. And beyond that there are very small cost savings, if any, to getting bigger. In the meantime, you're losing a lot of choice through consolidation, essentially putting yourself in a situation where there's only one employer of teachers in the county, where there's only one school district that can provide services to the children in the county and as a result when you have no options, you are a bit of a captive audience on the other side. That can lead to wasteful patterns of spending. We've done a lot of work with Texas data and what we found is where there is more choice, there is more efficiency, and that's a good thing for Texas. So I would advocate for consolidation amongst the very small districts where they are not- very sparsely populated communities. Small districts in urban areas don't have a necessity to be small and they lose a cost advantage. They're more expensive to operate. In rural areas, your problem is really about the lack of population density, about having to disperse the schools over a broad geographic area and even if you consolidated the districts, you wouldn't consolidate the campuses. And a lot of the cost savings from consolidation comes from consolidating the campuses.

Josh: Makes sense.

Rob Shearer [Director of Communications, Commit Partnership]: So like a San Antonio would probably lend itself to consolidation given how fractured the school districts are right in the heart of the city?

Dr. Taylor: Yeah, but San Antonio doesn't have very many districts in Bexar county that are below cost effective scale, and three of them are military bases. So it might make sense to consolidate them into some sort of Department of Defense organized, structured thing.

Rob: But folding Edgewood into SAISD-?

Dr. Taylor: Folding Edgewood into SAISD, there are very limited gains from economies of scale from that . And if you wanted to realize those gains, you would have to start consolidating schools, and closing schools, and that is a politically fraught thing to do.

Josh: Especially if you're consolidating schools across lines that have been in existence for a long time.

Dr. Taylor: In our report we identified exactly which districts in Bexar county are so small that it costs more to operate them then would be optimal. And then it'd be, that's one of the, I don't know, places where you could definitely make the argument for consolidation and you have- in most of the major metropolitan areas of Texas- still a lot of choice amongst effectively, efficiently sized districts. So consolidating them would just kill the choice, kill the discipline of the market without any gain in economies of scale.

Josh: Another one of the possible efficiency measures that you've written about is incentive pay, and specifically you mentioned this possible policy tension between incentives that are strong enough to elicit a behavioral response and the need for teacher acceptance, which is something that we're dealing with here in Dallas with our TEI incentive pay program. And so I was just hoping to hear you speak a little bit more on that tension and what you do with that.

Dr. Taylor:  So we did an analysis of a series of teacher incentive initiatives in the state, the Texas Educator Excellence Grants and the Governor's Educator Excellence Grants. These were incentive pay plans where the state was providing supplementary funds for incentive pay and the districts, and/or the schools were charged with designing their own incentive pay plans and this was a deliberate effort to get to greater teacher buy-in, to have the teachers design the incentive pay plans for themselves. And what we observed was they designed for themselves very egalitarian systems where just about everybody got a small amount of money. Basically, if you have a budget, if you have a million dollars to spend on this project. You can divide it up so that everybody gets a small amount or a few people get a large amount. What the teachers designed for themselves were small amounts where everybody gets some, rather than the equivalent outlay of large amounts that only a few people get. And so if the bonus award is going to be on average in the ballpark of a couple hundred dollars a month, at best, you know, how much behavioral change are people going to do chasing that kind of level of funding, especially if you don't understand exactly what it is you need to do to be the person most likely to receive the award. One of the things that characterizes education, not in all other working environments, is that the teachers frequently have a really good feel for who are the most effective teachers in the building. They know who amongst their colleagues are really effective and who amongst their colleagues are phoning it in and who amongst their colleagues, you know, should get some retraining. So if you told them that you had an award system that the best teacher in the school is going to win, we all go, oh, that's Mary. And so Mary knows she's the best in school, so she didn't have to work any harder. She's the best in school. Everybody else knows Mary's going to get the money. So no matter what I do, I'm not going to get the incentive award. So now these sharp incentives, big dollar amounts, but they're going to the best person in the school, we all know who that is already, not going to induce a behavioral change. So there's a lot of deep solid economics behind the optimal design of incentive pay plans and we just haven't gotten there yet to develop an incentive pay plan that's going to incentivize a lot of people to move and be as sharp enough, strong enough, big enough dollar amount to actually get somebody to credibly say, yeah, I'm going to retrain. I'm going to adopt this new practice. I'm going to try something outside of my comfort zone because of this reward. In other countries, we have been able to do these experiments where the awards were pretty substantial and the results are typically quite favorable to incentive pay. But in the US we tend to have relatively weak incentives or we have sharp incentives, big dollar amounts, but we all know who's going to get them. So again, nobody's motivated to change their behavior.

Josh: You've got to find a balance.

Dr. Taylor: There's a balance What the literature suggests to me, is that if you really want to do incentive pay, you have to go all in. It has to be a big part of everybody's compensation. Not windfall amounts that are relatively modest. And there has to be possibility for lots of people to be rewarded so that we're not going to just assume that whomever is already most talented in the building is going to take home the prize and the rest of us are kind of going to go be supportive.

Josh: Makes sense. So a while back, while you were still with the Fed-

Dr. Taylor: You did some ancient history work, didn't you?

Josh: Well, it's interesting because it's from a somewhat rosier time in school finance in Texas, and you wrote four principles of public school finance that are sort of the gold standard: treat equals equally, respect local tastes, match benefits with taxes, and avoid unintended consequences of redistribution. Do you feel like the school finance system as it exists now meets all or any of those ideals?

Dr. Taylor: Well, I think those are all aspirational ideals and I still stand behind them. Good economics is not always good politics. So there frequently are unintended consequences- or consequences that somebody intended, but it wasn't me- to policy initiatives. I believe that there are a lot of things we do right, but there are also a lot of things we could do better. And I think that one of the things we need to do better is to really resource the schools in an equitable way.

Josh: That makes sense. Can you go a little more into that?

Dr. Taylor: Oh! Deeper.

Josh: What that would look like, from an economist’s point of view?

Dr. Taylor: Well, from my perspective, that would mean that there would be more dollars in urban areas and in remote rural areas to reflect the differences in labor costs, that the current funding formula adjustments for economies of scale are probably over-generous to the midsize districts and insufficiently supportive of the smallest districts in sparsely populated areas. It is very expensive to operate schools in sparsely populated areas of the state. And I don't think the funding formula fully recognizes that. It's also very expensive to operate schools in high labor cost environments, and the cost of education index makes partial correction for that. But it's lost a lot of its match to what needs to be done. There should be a bigger gap between some of our urban areas and some of our sweet spot rural areas that are large enough to have the amenities of modern life and not be terribly remote but yet also have affordable cost of living and living standards that are attractive to today's teacher corps. If I were trying to redesign schools, I would do my darnedest to get away from teacher salary schedules that reward years of experience and advanced degrees since we have very limited research support for those metrics as being indicators of classroom effectiveness. Some people with a bachelor's degree are just incredibly gifted and talented teachers, and some people with a master's degree or PhD can't teach their way out of a paper bag. And then you have some people that get better over time. There's a lot of suggestion that a rookie teacher is learning quickly, but that they aren't as effective in their first couple of years as they’re going to be later in their career. But our salary structures tend to reward endurance in ways that don't always align with classroom effectiveness. So there's a lot of reworking that could be done in the way we handle labor costs in education.

Josh: I was also surprised to read from some of that Southwest Economy stuff.

Dr. Taylor: Yeah.

Josh: Talking about, and we're very much broadening the scope now from just the state level, but you were analyzing the PISA data, PISA. I had never even heard of before, but this international standardized test, and when you look at it from that level, you said specifically education funding and student performance, there is not much relationship between the two when you, when you look at the country level. So what do we do with that in terms of thinking about funding schools at the local level and the state level. And you even specifically talked some in that about how we have low classroom sizes with low salaries and if you pay teachers more and have higher class sizes, then that might be more economically efficient. But then there's also other data about how that might not be developmentally appropriate in certain instances. So how do we weigh these things? You know, there's just a lot of competing ideas out there.

Dr. Taylor: There are a lot of competing ideas and there is some confusion. I firmly believe that the educators are really doing what they believe will be the best thing for the students in their care. And the problem that you run into is that we don't share best practices very well from district to district. When a district figures out a strategy that works effectively for them, there's not a lot of transmission of that knowledge to other places. So we tend to do what we've done in the past. And so one of the things that I think is a useful tool for getting to more effective schools is just promulgating information about who's able to get to the performance that we want with the dollars that we already have in play. One of the most important things in getting school reform is to make sure people understand that it is in fact possible, that there's another school over here that is funded no better than you are and their kids are knocking the socks off of your kids. And so you should be looking to them for the techniques, the ideas, the tools. Some of it is very generalizable. They may have a curriculum model that's more effective. They may have a teachers specialization or not specialization, they may have adopted some of the best practices coming out of the charter schools, that have to do with an emphasis on everyone being capable of being college-ready and not accepting low expectations for kids because of their demographics.

Josh: So it's getting a little bit broader than just education. But of course it's all interrelated. You've also written about a number of economic issues. And so in looking for some additional sources of revenue to possibly address some of these issues in education, you've talked about some inefficiencies within our tax system in the state of Texas. And two things I'd love to hear you speak a little bit more on are a value added tax and “distortionary exemptions” in the sales tax.

Dr. Taylor: Okay. You're trying to get me in trouble, aren't you? So when an economist is approaching a question of revenue, okay, you essentially start from an understanding that one has made the case that revenue is required. Okay. Given that you've made the case that revenue is required, then you want to raise that revenue in a way that minimizes the distortions per dollar of revenue. So you want to have as a general rule, low tax rates applied to broad tax bases, and the value added tax and improvements to the Texas sales tax would both accomplished that goal of broadening the base so you could lower the rate and still raise the revenue that you need for the programs that you have justified meeting the revenues for. A value added tax is a tax on all consumption that people do, and only consumption. So you don't tax business to business transactions under a value added tax, but you do tax everything that people consume. If you tax everything that people consume at the same low rate, then the taxes don't distort the choices people make about whether or not I want to eat at home or eat in a restaurant. They don't distort the choices people make about whether or not I want to spend my money on a sweater or spend my money on a savings account, or do I want to buy now or buy later? Those decisions aren't distorted by the tax code, and anything that basically keeps the government out of those personal decisions, absent some, what we call externalities or special circumstances, is a desirable thing. So you want the tax code to be broad-based and low rate. If you tax everything that people consume and only things that people consume, then you're going to tax everything once rather than repeatedly. And you're going to be able to apply a relatively low tax rate and raise the revenue that you need in a minimally distortionary kind of way. One can get to a value-added tax or a consumption tax using a sales tax as the model. Basically what you do is you would tax everything that people consume and you would not subject business to business transactions to a sales tax, and that leads to a lower, more uniform, fairer tax rate than a world in which you tax business to business transactions, because the businesses pass it along to their customers. And you'll have a situation with a firm that is purchasing a lot of services in the market and paying taxes on their purchase of accounting services and their copying services and their legal services and their raw materials. And if they pay tax on each of those, then each of those taxes gets rolled into the final price to the consumer. And in some sense the good can be taxed many, many times at different stages of the production process. More and more taxes layered on. With a value added tax, you only tax at the final stage, the final sale to the consumer. And as a result of taxes are uniform, and because they are more uniform, they're less distortionary. The Texas sales tax has a number of features that make it far from an efficient form of taxation. There are a number of things that consumers buy that are not subject to the sales tax. The tax falls most heavily on goods but not on services. And that's problematic because a lot of households, particularly more affluent households purchase a lot of services, and so when you tax goods, you tend to tax regressively, you tend to tax more heavily on low income households. The tax exempts food from the sales tax out of an admirable effort to try and protect the poor from the burdens of the sales tax. But the problem is that Ross Perot's filet mignon is also tax free. And it would be more equitable and would allow us to raise the revenue we need at a lower rate if we were to target individuals for the tax exemption, not products for the tax exemption. And so you should broaden the sales tax base to cover everything that consumers buy. Only things consumers buy, get rid of the business-to-business transactions there. You could raise similar kinds of revenue with much less distortion, much less government-driven decision making on the part of the business community.

Josh:         Awesome. So last thing, and again, I realize it's somewhat speculative and you don't want to get in trouble. You were very politic to preface that by: You’ve got to make sure the revenue is required before you go looking for it, which makes sense. And so assuming that a lot of these measures that we've talked about in terms of ending inefficiencies in how we finance schools, say I email all of these essays that I've been looking over to all the legislators. They're like, "oh, how have we been foolish?"

Dr. Taylor: Yes! Eureka!

Josh:          We gotta revamp the cost of education index, we get a comparable wage index, incentive pay, all this stuff. We clear up all of these inefficiencies. Do you- on the other side of that, do you think then that we would no longer need any additional revenue, or that there is still the possibility that even should we completely optimize the efficiency there's going to be a gap?

Dr. Taylor: Well, revenues will be needed. How many- whether we need more than we have now, depends enormously on what we expect the school to accomplish. What are the outcome goals?   The higher our standards and expectations, once you bring the inefficiency out of the system, the higher the standards and expectations the more money it's going to take. The challenge right now is that a lot of folks, myself included, don't have great confidence that putting additional resources into the system won't leak out of the bucket rather than going into improve student performances. So we want to fix the vessel before we fill 'er to the brim.

Josh: To say that it needs fixing, which I think we would all agree with, is not necessarily to say that then it wouldn't need additional water to fill that bucket, but the priority has to be on fixing it first.

Dr. Taylor: Well, I think that you could actually pursue some of this in parallel, because our work over the last couple of decades has been looking at situations where there are clear resource needs that the state should address. Even in the most lean and economic and efficient modeling of cost of education work, there were still districts out there that needed more resources than they currently have to be able to accomplish the goals we have for them. And I don't want us to let the goals slip, so I've got to recognize that more resources are needed there. There's only two ways to get more resources there. One is to bring more resources into the education system. The other is to reallocate existing resources from one district to another. And I'm not sure the political will is there to do the reallocation. I want to get deep into the analysis of Texas. Basically, the work that we did on the consolidation study you looked at could be expanded and interpreted in the context of an estimate of the cost of education for a variety of performance standards.

Josh: How long does something like that take?

Dr. Taylor: The consolidation study was three months. It depends on whether or not we are teaching at the same time.

Rob: I did have one other question. So you mentioned, it doesn't seem like additional funding necessarily moves the needle on outcomes for students. There's not a tight correlation there, right?

Dr. Taylor: There is a correlation there once you control for inefficiency, but there's substantial inefficiency in the system.

Rob: So I wonder if, similar to our earlier conversation about if you take teacher incentives and spread them across all teachers, you don't end up with a behavior change, I wonder if part of the struggle we have with school funding is we tend to do the same thing. We take equal amounts of money and apply them roughly to all of the same students. For example, in Texas we've got a huge percentage of students learning English as a second language. We have the second lowest investment in the country for ELL students, and I can't help but wonder, are there student populations that we aren't investing enough in to get equal outcomes? I believe that talent is distributed equally in the state of Texas. Opportunity is not.

Dr. Taylor: Right. I think though, that it's really important to recognize that there are economies of scale in ESL and bilingual instruction as well. If you have a student that needs bilingual instruction, there are only 15 of them in the elementary school, then you're bilingual instruction class is going to be 15 kids to one teacher, that's expensive. Twenty two kids to one is a much more cost-effective size. So there are reasons to believe it's cheaper to operate a bilingual ESL programs in Texas with its large fraction of kids eligible. We got critical mass in many of our districts in need of the services. And so I would expect us to spend less per pupil in the program than other states do, where they are treating the kids in borderline one on one tutoring kind of contexts. Some research out there suggests that if all of the ESL or bilingual education kids are sharing a common language at home, it is easier and therefore more cost effective to operate the classroom then if you're trying to juggle 15 home languages. And last number I saw was something like 98 percent of the ESL students in Texas are Spanish speaking in the home, which gives us a cost advantage in delivering bilingual education or English as a second language in Texas, because we're delivering it to a Spanish speaking population by and large. And so we can get to some of these economies of scale at the classroom level in delivery of services to those kids. So I would be a little queasy about saying that the state spends less per bilingual kid and we have so many more of them. It might be that we spend less because we have so many more of them, it allows us to do it more cost effectively than say a North Dakota could ever hope to do with their bilingual education kids. The other thing that we didn't get into that's probably relevant here is that we do a really poor job of identifying a student's socioeconomic status and poverty. We use the same income threshold in Texas as they do in New York City. And so the kids in poverty in some parts of the state are much more desperate in their home circumstances than the kids identified in poverty and other parts of the state. Where labor costs and housing cost is low, you will have intact, two parent, high school graduate households below the poverty threshold. In other parts of the state it's much more severe home circumstances to be eligible for free and reduced price lunch. And when you don't measure poverty well, you don't target your compensatory education resources well, and as a result, you got a lot of leakage out of that bucket. So if I were trying to fix two big pieces of the funding formula, one would be the cost of education index, and the second would be how we measure poverty. And then the third would be scale.