In Census Finance Data, Most Charters Are Not Quite Public Schools

Last month, the U.S. Census Bureau released its annual public K-12 school finance report (and accompanying datasets). The data, which are for FY 2009 (there’s always a lag in finance data), show that spending increased roughly two percent from the previous year. This represents much slower growth than usual.

These data are a valuable resource that has rightfully gotten a lot of attention. But there’s a serious problem within them, which, while slightly technical, hasn’t received any attention at all: The vast majority of public charter schools are not included in the data.

To gather its data, the Census Bureau relies on reporting from “government entities." Some charter schools fit this description neatly, such as those operated by governments or government-affiliated bodies, including states, districts, counties, and public universities. But most charter schools are operated by private organizations (mostly non-profits), and finance figures for these schools are not included in the report (the Census classifies them as "private charter schools").

What does this mean? Well, for one thing, it means that the overall spending figures (total dollar amounts) are a bit understated. Charters only account for a relatively small proportion of all public school enrollments (around 5-6 percent); still, given the huge amounts of money we’re dealing with here (the U.S. spends roughly $600 billion a year), we’re talking about quite a bit in absolute terms. Perhaps more important is the potential effect on per-pupil spending figures – the way that education financing is usually expressed.

Per-pupil calculations allow one to account for variations in enrollment among years and locations. For example, some districts/states have more students, so of course they spend more – you need to see how much they spend while at least crudely “controlling” for how many students they serve, and per-pupil figures help to do that (though they don’t account for differences between students in terms of income, disabilities, etc.). The same goes for spending in the same state or district across years – if district/state enrollment goes up or down, spending will too. Comparisons over time thus require that we account for these changes by dividing spending by the number of students.

But even per-pupil estimates are subject to potentially serious bias, due to the charter school exclusion. This is, in many places, a problem at both the state  and district level. First, at the state level, if the excluded charters spend more or less than regular public schools and included charters, then this could skew the per-pupil amounts, making them appear lower or higher than they really are (especially in states where there are enough charters to affect the overall average). In other words, if there are substantial spending differences between the excluded charters and the regular public schools and included charters, we might not be getting an accurate picture of how much each state spends on public education.

Now, one might argue that the state figures are not necessarily accurate, but the district spending figures should be, because most charter schools (whether or not they are included in the Census data) are districts unto themselves. So, even though we might not know for sure how much entire states spend per pupil, we can still get accurate estimates for each individual district – a district spends what it spends, regardless of whether or not there are charters in the area, so the district-level data are accurate.

This is not quite true, of course. Many districts spend money on their “host” charter schools (the ones geographically located within the district), even though the latter are not technically part of the district. For example, many school districts pay the transportation, food, and other costs for the charter schools within their boundaries, and these expenses are sometimes substantial. As a result, in the Census data, school districts are being “charged” for services that are actually being used by charter school students who are excluded from the finance data. This inflates the per-pupil spending amounts for regular public school districts, and has the opposite effect on charter school spending. This type of problem persists even when looking at district or state spending across years. If the number of excluded charter schools increases or decreases, this would create bias in the trend as well.

(Note that these issues, at the district-level, do not stem from the exclusion of charters from the Census data, but rather from the manner in which many states and districts track their spending – they don’t “follow the money” that is spent on charter school students.)

To be fair, some systems “follow the money” better than others.  For example, a number of states fund their charter schools directly (rather than making the money pass through districts first).  And this represents yet another source of charter funding-related bias in the data – the differing features of various finance systems.

It would be difficult to solve these problems at the district level (you’d have to design systems to follow the spending on each student, which some systems do), but there is an easy state-level solution: Include all publicly-funded schools, charter or not, in the finance data. Now, by “easy," I mean that it is easy to say, but not necessarily to do. The Census Bureau is constrained in terms of their data collection, and I’m guessing (though I’m not sure) that its ability to collect all the data would require some kind of state legislation requiring all "non-public" charter schools to report their finances to the Bureau.

In the meantime, we need to be careful about interpreting these Census data. For example, the overall amounts of U.S. spending on public education are simply not accurate, as they exclude a fairly large number of schools that receive public funding. Even the per-pupil estimates, for districts, states, and the nation as a whole are potentially biased, both statically and over time.  It’s very difficult to untangle all the problems, or even to estimate the amount of bias in the data.  Such efforts will have to proceed on a case-by-case basis (individual states or districts).

More generally, there will always be some inaccuracy in finance data at every level, but given how much attention we all pay to public school spending, it would seem important that our data at least not exclude a large and growing subset of public schools, especially when that exclusion likely causes serious inaccuracy in the estimates. If charter schools are public schools, then their finance data should be reported in our national estimates, just like everyone else’s.

- Matt Di Carlo