The spin is on. As soon as the annual school district level U.S. Census fiscal survey data are released, news outlets across the country take their shot a spinning the data to show just where their state stands. New York #1! Utah… dead last! Hawaii “above average.” Spending just really high (totally out of context)! Typically, news outlets point out spending is high when they wish to argue that it’s too high… and we should do something to curb it. No mention is made of outcomes achieved with that spending, or which districts in the state are responsible for the high average. When spending is reported as low, the spin is generally that it is too low, and that state policymakers should do something about it.
Allow me to briefly present a slightly more nuanced picture. For the past few years, and in a number of publications, I have used a statistical model of the national school finance data to correct for such issues as a) economies of scale and population density, b) regional variation in competitive wages, and c) variations in student needs. I use this model to project what a school district, with comparable characteristics, would have in state and local revenue per pupil in each state. The methods of this madness were used in this study: http://epaa.asu.edu/ojs/article/viewFile/718/831
Here are some of the results with the 2007-08 Census Fiscal Survey data (with the model built on data from 2005-06, 2006-07 & 2007-08).
Before getting to the modeled estimates of comparable state and local revenue, lets take a quick look at the relative educational effort of each state, or the combined State and Local Revenues for K-12 education as a share of Gross State Product. Vermont and New Jersey lead the pack on this on, with other states including Maryland and New York in the mix. Note, however, that this effort can be quite unevenly distributed. In fact, it may be the case that a significant amount of effort is going into local property tax revenues being raised by the richest communities in a state. Yeah… it’s still a lot of effort, but selectively distributed among those who can put up that effort and choose to as long as it benefits (or is perceived to benefit) their own children. Total effort provides a limited window, but important one nonetheless.
Fun Fact about this first table – TAKE A LOOK AT OUR RACE TO THE TOP, ROUND 1 WINNERS! (47TH & 50TH ON EFFORT!!!!)
Now to the model based estimates of who’s really in the top and bottom ten on state and local revenue per pupil for elementary and secondary education. Let’s begin by looking at those states where the lowest poverty districts have the highest and lowest resources.
Yep, New York is #1 in per pupil state and local revenues for very low poverty districts! Indeed, very affluent Long Island and Westchester County school districts in New York State spend about as much as any districts in the nation, largely because they have the financial capacity to do so (and partly because the state has enabled them to!)
Next in line in funding for very low poverty districts are Wyoming and Vermont, which really don’t have many children attending incredibly high poverty districts. Notably, New Jersey falls well behind New York state for low poverty districts, and many of New Jersey’s affluent suburbs lie in the same labor market with the higher spending affluent New York suburbs. And then there’s Tennessee – one of our great RttT winners. Of course, as I have shown on a previous post, this works fine for TN, which as the lowest state assessment cut scores – so most of the kids pass the tests anyway (low standards & low funding – a winning combination indeed)!
The next table ranks per pupil funding for high poverty districts. Notably, New York is NOT in first place on this one. New York drops to 6th, but the situation is somewhat more complicated. While this might appear okay, it can be particularly difficult for high poverty New York state school districts to recruit and retain high quality teachers when they are surrounded by so many affluent districts which already hold the recruitment and retention advantage, and have substantially more resources. For high poverty districts, New Jersey and Wyoming come in first. Wyoming is simply high across the board. And yep… there’s Tennessee again – our RttT winner in 47th place!
This next table ranks the within-state FAIRNESS of the state school funding distribution – where fairness is determined by taking the ratio of high poverty funding to low poverty funding – with the implicit assumption that state school finance systems should provide for additional support in districts serving children with greater needs. Now, this table must be taken in the context of the previous two. For example, Utah comes in first on “fairness.” But, in this case, this merely means that low poverty districts in Utah get nothing, and high poverty districts in Utah get next to nothing! In a twisted sense, that’s “fair?????”
Among states not at the bottom in overall resources, New Jersey, Ohio, Minnesota and Massachusetts seem to be driving additional resources into higher need, higher poverty districts.
States at the other end of the spectrum include New York, Pennsylvania and Illinois. These are among the historically least equitable large, diverse states in the country. Now, to Pennsylvania’s credit, these calculations precede the phase-in of their new funding formula which the governor has continued to support even during the recession. New York and Illinois are another story. Yeah… New York also implemented – okay – kind of planned to implement a new formula. That didn’t get very far, and it is highly unlikely (okay, almost entirely unlikely based on other analysis I’ve conducted on more recent NY data) that NY has actually improved since 2007-08. Illinois hasn’t even tried – in fact, Illinois just keeps getting worse and worse!
Now for an obligatory point – Many argue that the overall funding level in states is simply a function of their wealth. Wealthier states, like wealthier school districts within states simply have the ability to spend more. That is indeed partly true. But effort also matters – remember that first slide above? This scatterplot shows the relationship between state effort and funding levels in a hypothetical average poverty school district. There’s actually a reasonably strong relationship here, but for a few quirky outliers. In fact, based on additional analyses, a state’s effort explains about as much of the funding level as does a state’s wealth.
So, Mississippi is a very poor state that puts up relatively average effort, but simply can’t get very far with that effort. By contrast, Tennessee and Louisiana both have much higher fiscal capacity (measured by gross state product per capita) than Mississippi, but they simply don’t use it. Tennessee has little excuse for its spending level! Nor does Louisiana!
Finally, here’s a snapshot of the association between 8th grade reading and math NAEP performance and funding levels across states. As it turns out, funding levels for high poverty settings were most strongly associated with NAEP performance for all students. As one can see, there exists a reasonable correlation between funding levels and NAEP mean scale scores. That said, as I have noted in previous posts regarding such relationships, there’s a lot of circular stuff all tangled up in here. Wealthier states with more educated adult populations supporting higher education spending – and supporting and encouraging their children to do well in school, etc. But, it is difficult to conceive how a state in the bottom left corner of this picture (very low funding in high poverty districts – and most likely, low funding across the board) can begin to lift itself out of that corner – or Race to the Top. Financial resources are a necessary underlying condition, albeit easier to achieve in some states than in others.
Note: Difficulties arise when trying to make simple comparisons of funding levels and funding gaps with achievement gaps between poor and non-poor children in each state a) because income thresholds used for subsidized lunch status characterize very different populations from one region of the country to another and from rural to urban settings within states, and b) because gaps between non-poor and poor children in states depend significantly on how wealthy are the non-poor and how poor are the poor. Sadly, these complexities make it very difficult if not impossible to use NAEP data to untangle the relationship between funding differences between lower and higher poverty districts, and outcome differences between children attending those districts in different states:
I discuss the poverty measurement problems here:
Kevin Welner and I discuss evaluating the relationship between state school funding distribution and student outcomes here: