This is the model for new school construction. (HT: Arkansas Project)
I understand why House Democrats included $20 billion for school construction in the $819 billion stimulus package they passed last week. They need to throw money out the window as fast as the printing presses can make it. That way they can say they are doing something to “help” the economy. And as long as they are doing something they might as well help their friends in the educational industrial complex.
What I don’t understand is why some normally smart people feel compelled to justify this school construction spending by claiming that it will help students or that it is badly needed. There is no evidence to support these claims. That’s why I was surprised to see Sara Mead, senior research fellow at the New America Foundation, write:
“But perhaps the most important reason to invest in school construction is that our students need it. Just as Americans have underinvested in our bridges, roads, and other infrastructure, we’ve also underinvested in our education infrastructure. The American Society of Civil Engineers Infrastructure report card gives our school buildings a grade of D — lower than grades for bridges, rail, or public transit infrastructure. According to the National Center for Education Statistics, it would cost $127 billion just to renovate and repair crumbling or outdated school facilities to good condition. Poor school facilities don’t necessarily prevent students from learning, but, it’s unconscionable that we currently ask students to learn, and teachers to work, in buildings that are overcrowded, inadequately heated and ventilated, poorly maintained, and in some cases literally falling apart. The contrast between schools and other buildings sends our most disadvantaged children a devastating message about the value we place on their education.”
As readers of JPGB may recall, the research literature finds no relationship between school facilities and student achievement (outside of developing countries, where a grass hut and mud floor may be a hindrance in bad weather). To repeat: “In the Handbook of the Economics of Education, Eric Hanushek reviews all of the research meeting minimal quality standards regarding the relationship between school facilities and student performance. He identifies 91 analyses on the issue in the U.S. and finds that 86% of them show no statistically significant relationship. Of the remaining 14% of analyses that did show significant effects, 9% were positive and 5% were negative. ”
If the evidence generally fails to find that better school facilities improve student outcomes, it’s not clear how Sara Mead can support claims like “our students need it” or that current facilities send “a devastating message about the value we place on their education.” If it has no effect on their learning, how devastating could it be?
Perhaps Sara Mead relies upon the assessments from the American Society of Civil Engineers (ASCE) and the National Center for Education Statistics (NCES) to support her claims. Let’s look into those references a little further to see whether we should believe them or believe the research reviewed by Rick Hanushek.
The ASCE grade of D- for school infrastructure is in a report card issued by the interest group representing engineers who design public infrastructure. To determine the grades ASCE “assembled a panel of 24 of the nation’s leading civil engineers” who then reviewed other reports and surveys of civil engineers. I would be more specific about their methodology, but they provide almost no information about their methods or standards. Their entire description of the methodology consists of 5 sentences. You can read it yourself to see.
In the 2009 report card they actually gave school infrastructure a D. But don’t worry, that is the average grade they gave to the 15 categories of public infrastructure they graded, which ranged from C+ to D -. They also determined that we need to spend $2.2 trillion (with a T!) on public infrastructure. How exactly did they determine that? Again, the five sentences describing their methodology didn’t exactly explain all that. But since they are a lobby group representing engineers who work on public infrastructure projects, I’m sure we can just trust their expert judgment that we urgently need to spend $2.2 trillion on what they do.
Using the NCES to support a need for $127 billion in school construction spending is hardly more persuasive. To be precise, the NCES did not determine the amount that needs to be spent on schools. They just surveyed school district officials and in the survey asked the school officials how much money they thought needed to be spent on their schools to bring them to “good” condition. The NCES does not represent this result as their own opinion; they portray it as the opinion of the people they surveyed.
Again, asking people who are the potential beneficiaries of public spending whether they need more money without any standards to define their “need” is unlikely to be a reliable method.
To illustrate the point, I’ve conducted a survey of “experts” on JPGB – the regular contributors — to determine the resources we would need from the federal bailout to bring the blog “into good overall condition.” One respondent identified the need for martinis, cigars, and a few Vegas junkets to improve our creative process. Another identified the need for a board game marathon only interrupted by Indian food buffet binges to sharpen our intellect. And another agreed with the first two and added the need for a spaceship, pony, and a lifetime supply of spicy wings and candy. (You can guess who’s who.)
I think I’m going to trust Hanushek’s assessment of 91 analyses that meet social science standards over the self-serving assessments of school officials and and a construction lobbying organization. If you disagree, then I expect that you’ll support our demands to add JPGB to the bailout since, after all, the “evidence” clearly demonstrates our need.
Jay P. Greene's Blog 
I wouldn’t argue against the need for proper justification… but I would consider tax capacity to be an issue for many communities. The less they have to spend on schools, the more they can spend on academics.
In our part of the country — Texas — the wave of schools came about the time of the wave of interstates. So we have many campuses now that have reached the end of their useful life in terms of HVAC and wiring, et al. So it would be, frankly, easier to use stimulus dollars than to try to convince taxpayers to pay for something that is not going to be shiny and new.
Stimulus dollars are also taxpayer dollars. The only difference is the taxpayer is less aware of the fact that that the federal dollar is taken from him/her and less aware of where it is being spent. I suppose the former makes it easier to take the dollar (if you favor that), but the latter makes it harder to ensure that the dollar is used well. You may think your local schools need new HVAC and wiring, but the federal program may well spend those dollars on different construction projects. If you raise the tax money locally, it may be harder to do, but it is more likely to go to where you think it is needed.
I agree that the $20 billion seems to have been allocated to school infrastructure rather hastily. Yet there are a few ways these improvements to infrastructure could possibly lead to improvements in student achievement. One way is by improving infrastructure that will free up money in the future. This is especially true if the money to be spent on projects that have future savings. For example investing in 10K in improving heating, air conditioning, pluming, electrical and insulation could be turned into 20K of savings over the subsequent 10 years. Those savings could be used to really try and improve student achievement. Another possibility is that structural investments can be used to as a way to attract and retain the best and brightest teachers. The private sector often compensates employees not just financially but also with better work settings, executive chairs, corner offices, private bathrooms, maybe there is a way where this practice can be exported into public schools.
Such ideas will probably only work in a more decentralized school district but and might not be possible in today’s large urban districts, but in principal I believe the idea has merit.
Tola, the question is – since extra spending on infrastructure seems to have been consistently wasted in the past (see the research Jay is citing), what reason do we have to think that future spending on infrastructure will be used productively? It’s easy to come up with hypothetical theories for how more spending might concievably improve student outcomes on the assumption that once the money disappears into the system it will be used that way and not diverted to feed the gravy train. But that’s the whole issue, isn’t it?
Martinis, cigars, Indian food buffets, and board game marathons?! Where do I sign up?
I’m trying to get my head around the claims made by Greg (Hi, Greg) and Jay here. The claim seems to be that a meta-study that admits that spending on facilities is a “poorly measured” variable in the underlying studies should be taken as evidence that school infrastructure spending does not contribute to educational outcomes. Looks a bit like a GIGO problem to me. At any rate, it’s hard to believe that school districts where buildings are dilapidated (in ways that students notice) don’t have a signaling problem on their hands.
And really the question is more the following: do we want to be the kind of people who allow our schools to fall apart? If you want to privatize the system, then you don’t have to care too much about that question. If you like public schools, however, it’s a more important question.
Brett, I don’t have too much to add to Jay’s response as far as the substance of your question goes. The studies do show quite clearly that merely increasing the overall spending level on infrastructure doesn’t lead to improvements. Obviously if we were doing something other than merely increasing the spending level – if we were introducing some new system that would spend the money differently – then we’d have to have a conversation about to what extent this evidence was relevant to that. But since the proposed policy is precisely to just increase spending with no systemic change that might give us some reason to think the money will be spent better this time than it has been every previous time, I think the real GIGO here is on the policy side: the longer we follow the garbage policy of throwing more money at the problem, the longer we’ll continue to get garbage outcomes from it.
But I also wanted to respond to your framing of the issue. You say “really the question here is … do we want to be the kind of people who allow our schools to fall apart?” No, it’s not. None of us is that kind of person. Really the question here is, do we want to be the kind of people who make policy based on the empirical evidence about which policies produce the best outcomes for students, or do we want to be the kind of people who make policy based solely on which policies make us feel good about the kind of people we are?
Hi Brett. Thanks for the comment. You are right that the studies Hanushek reviews focus on spending on facilities, which is not the same as the quality of facilities since the money may not be spent wisely. But since the proposed policy here is to spend more on facilities, the studies Hanushek reviews are right on point. Facility spending from the stimulus is no more likely to be spent wisely than past facility spending, so we should have no reason to believe that this new batch of facility money should produce desirable results.
Of course, I don’t want schools that are falling apart, but I also don’t want to pursue a solution that will fail to solve whatever problems exist. And perhaps school facilities are generally not too bad relative to what we really need from them, which could be another reason that additional spending fails to improve achievement.
Help me out here (if you’d like). State A does a study and finds that it really should spend $200 million immediately to keep its schools “safe, dry and healthy” or however they describe the priority level one projects like fixing leaky roofs and faulty electrical systems or crumbling structural concrete. (State A would be a small state if that were its analysis, as I understand these things. A few years ago, it would be a state that Jay knows quite well!)
Does Hanuchek’s review of studies that “poorly measure” something called “facility spending” tell us that state A should not spend the $200 million, or that if it does spend the money, it is acting “without foundation of evidence?”
Now if state A can’t afford to spend the money this year because it has a budget deficit, and the feds say, “apply to us, you can spend federal money on this as long as you weren’t going to already,” does Hanuchek (or any other study) say that the feds are acting “without foundation of evidence” if it sends the money to state A?
I understand the general point that you don’t trust schools to allocate money to projects that improve student outcomes, perhaps because schools are politically unable to do so or because they don’t have the institutional capacity to affect student outcomes absent some market-based reform.
I’m still not sure how that general critique morphs into a critique of a program designed to fix leaky roofs, given the fact that there will be no market-based reform that you would be happy about. In the second-best world of the institutional status quo, do you support federal help in fixing leaky roofs?
Of course, if you like charter schools, they can fix their leaky roofs under the stimulus plan, too, if I understand it correctly.
Maybe you’re saying that the bill, as written, allows spending on more than leaky roofs. Would a more tightly written set of requirements satisfy you? If so, why wouldn’t a bill with those requirements also cut against your reading of Hanuchek?
Brett, you treat the state’s so-called “study,” which (as Jay has documented) has no scientific validity, and Hanushek’s review of the findings of a large number of scientifically rigorous studies as though they were of equal weight.
You are misusing the “poorly measure” quote. The data in the studies Hanushek reviews do not “poorly measure” the impact of infrasturcture spending. They’re a perfect measure of that. They “poorly measure” the impact of infrastructure quality.
If this kind of construction binge is going to raise student outcomes, three things need to happen:
1) The legislature raises infrastructure spending.
2) The school system uses the money in an effective way, to build better infrastructure, rather than diverting it to feed the gravy train.
3) The improved infrastructure improves student outcomes.
Now, if you want to find out whether #3 will occur once you have already achieved #2, then the data Hanushek is looking at will be of limited value, because they “poorly measure” infrastructure quality.
However, if you want to find out whether simply doing #1 and nothing else will, by itself, lead to #2 and then #3 happening, Hanushek’s data leave no reasonable grounds for doubt that the answer is no.
What the state is proposing to do is simply #1 and nothing else.
“One respondent identified the need for martinis, cigars, and a few Vegas junkets to improve our creative process.” – That would be Matt.
Speaking of fixing stuff. I met with an engineer here in Vegas that won a contract to repair 82 problems in a school. The next year the school district requested proposals to fix an identical school with identical problems. He submitted a bid, including blue prints from the identical job he did the year before and proposed he could fix the problem for $900,000 since he had already done all the brain work.
The school district awarded the contract to another firm for $2.2 million – that firm had to reinvent the wheel.
By the way, Nevada has the 3rd highest capital outlays per student in the country…we are pretty close to the bottom in achievement too.
Hey Greg: in Hanushek, 2006: 890, he writes: “The studies that include measures of facilities or administration fail to show much of significance. These factors are generally, however, quite poorly measured – thus complicating the interpretation.” The same description of the underlying data (as opposed to the underlying spending decisions or the causal mechanisms) appears in Hanushek, 1997: 143 (“The measures of other school resources also are frequently measured poorly and tend to be available only at the district level”). If he’s making a quality / quantity distinction about choices among resource allocations, I don’t see it there.
We went to graduate school together, so you can guess my intuitions about the “scientifically rigorous studies” claim, particularly where we are talking about linear regression models viewed from 50,000 feet, in the aggregate.
I gather that your answer to my questions above is: ignore the school, since its budget request isn’t reliable. So should the baseline be zero new dollars? Or zero dollars for fixing roofs entirely?
I’ve hijacked the thread enough and will let you respond. We can take it offline if you’re interested.
Brett, you are raising excellent points. If Hanushek fails to find a significant relationship between facility spending and student achievment only because we cannot properly measure the spending allocated to each building, then it is conceivable that such a relationship exists but we were unable to detect it properly. While I think this is possible, it is also unlikely. The measures in the studies Hanushek reviews are imperfect, but they are not complete junk. District facility spending and building facility spending are obviously going to be related.
Besides, the burden of proof should be on the people who want to spend $20 billion more. If we are going to do that, we should have some good evidence to suggest that it will be worthwhile.
And the suggestion that we could reduce facility spending to zero is not reasonable. Similarly, Hanushek finds no relationship between spending in general and student achievement, but that does not mean that total education spending could be reduced to zero without harm. It just means that the marginal return on additional dollars is zero. We don’t know at what point reductions would become harmful but it is certainly above zero.
Brett, it’s not hijacking the thread if you stay on topic, which you are doing.
Now, if you’re really going to wave your hand and say that you don’t believe in any claims to scientific rigor on this subject, such that the transparently bogus propaganda of a government monopoly that diverts billions of dollars annually from legitimate public use to the private benefit of its constituents, destroying the lives of millions of children in the process, has equal probative weight with the careful work of serious scholars whose analyses have been exposed over many years to the critical review of the whole scientific community and have been pretty universally agreed by the profession at large to be reliable, well then, I have no more to say. But if you’re willing to admit that you’re not really serious about that, read on.
When Hanushek says facilities spending is “poorly measured,” I take that to mean three things: 1) while we know how much money was reportedly spent on facilities, we don’t know what that money was actually spent on, since the system is notorious for its “sloppy” (read: corrupt) bookkeeping; 2) even if we took it for granted that the money that was reportedly spent on facilities was actually spent on facilities, we don’t know exactly what kind of facilities (i.e. good or bad) it was spent on; and 3) the data do not allow us to perform building-by-building analysis.
When you say that Hanushek is questioning the quantity as well as the quality of what was spent, I take you to be pointing to meanings #1 and #3. And I can see that above I implied that I was only thinking about meaning #2. That was, indeed, what I was mainly thinking about at the time. But the point I was trying to make – raising the infrastructure spending level in the systemwide school budget doesn’t mean the money will actually go to improve infrastructure quality – really covers all three of those problems. An overall increase in the infrastructure budget won’t necessarily produce better infrastructure because 1) it’s ridiculously easy to divert the money to other uses besides infrastructure and never get caught, because nobody ever checks; 2) it’s ridiculously easy to spend the money on infrastructure that doesn’t contribute to school quality, because in most states there is not much effective accountability for school quality; and 3) the state legislature, operating at 50,000 feet, can’t control (and in fact it doesn’t even know) which buildings the spending will go to, and thus nobody can stop school districts from diverting the spending away from schools that need it and to the schools that are more politically connected or whatever.
Perhaps it would be clearer to draw a further distinction in the three-step process I described above, a distinction between the amount of money the state believes is being spent on infrastructure and the amount that is actually spent on infrastructure. So for an increased-spending-on-infrastructure scheme to work, four things would have to happen:
1) The state raises the amount budgeted for infrastructure.
2) The additional funds budgeted for “infrastructure” are actually spent on infrastructure.
3) The additional spending on infrastructure is carried out in an effective manner and actually raises infrastructure quality.
4) Improved infrastructure raises student outcomes.
Now, when I said Hanushek’s data are a perfect measurement of the spending amount, what I meant was that his data are the exact same data that the state has under its control when it makes a policy decision to increase infrastructure spending, i.e. the amount of money written down in the official 50,0000 foot budget under the heading of “infrastructure.” My point was, there’s nothing the state (or the feds) can do to ensure that increased spending will be wisely spent.
Before, I only pointed out that infrastructure spending is often carried out poorly. You have added the even more important point that the system is in fact so thoroughly incompetent and/or corrupt that it can’t even be counted on to actually spend the money in the “infrastructure” category on infrastructure. I don’t see what else it can possibly mean to say that the amount of infrastructure spending is “poorly measured,” i.e. the amount actually spent on infrastructure doesn’t match what the system’s budget reports was spent on infrastructure.
The real control over how money gets spent at the level of the individual school building is so far out of the state’s control (never mind the feds) that the state can’t even reliably measure what is being spent on what at that level – and what you can’t measure, you can’t control.
So you can see that, far from weakening Jay’s conclusion, your points only strengthen it.
I’m happy to accept the correction – thanks for setting me straight! 🙂
[…] high school is adequate and there was no need to demolish it entirely and replace it with a new Taj Mahal. Besides, there is no evidence that fancy buildings improve education. Buildings don’t […]
[…] new $578 million high school in LA Unified School District. As we’ve written before on JPGB, buildings don’t teach kids, people do. Given the way school districts squander their resources, maybe they’ll soon need another $26 […]