Why Critics Should Think Twice Before Assuming Iowa’s ESA is Causing Tuition Hikes

A new study in the Annenberg Institute’s working paper series purports that Iowa’s new education savings account (ESA) led private schools in the state to raise tuition prices. Social media commentary about this paper has been disappointing. While many are looking for any opportunity to say something negative about school choice programs, this working paper currently lacks the scientific rigor to conclude anything at all.

There are three major reasons to be skeptical of this paper and its conclusions, one analytical and two methodological.

1. SCHOOL CHARACTERISTICS MATTER.

Two school characteristics are crucial when discussing private school tuition: how much they were charging before the ESA, and how close they were to capacity.

There are three kinds of private school tuition rates in Iowa: those charging less than the ESA amount, those charging about the same as the ESA amount, and those charging more. It matters greatly how tuition increases vary across these three categories. Iowa’s ESA program dictates that students can receive 100% of the value of base state per-pupil aid, or about $7,400. (This is about 62% of total per-pupil public spending, which includes other revenue sources.)

If the increase in Iowa’s statewide average private school tuition is driven by schools that were charging less than the ESA amount, then accessibility does not change. If a school was charging $5,000 per school year prior to the ESA, and $7,000 per year after its enactment, even students who could not afford the lower tuition would be able to attend at the higher rate when utilizing the ESA. On the other hand, if unusual tuition increases were mostly occurring among schools already charging more than $7,400, raising rates far in excess of the ESA allotment, then yes, those schools are keeping themselves inaccessible to some students.

Yet the researchers did not make this distinction, even though they should have all the information they need to understand where tuition increases are occurring. Considering that most schools in their dataset charge less than the ESA amount (Figure A1 in the paper), it seems very possible that the tuition increases are driven by lower-cost schools, whose price increases generally do not affect accessibility. Perhaps the researchers will dive into that in future editions of this project.

Capacity must be considered as well. A school with many open seats will look at tuition differently from a school at capacity. The school with open seats may wish to keep tuition down to remain competitive and attract more students, which could increase total revenues more than just raising prices on current students. But a school at capacity has no need to attract more students and could perhaps afford to charge its current students more. So far as I know, no research exists in the elasticity of demand for private education in Iowa, nor the share of Iowa private schools at capacity. But this would be very useful information to understand how schools respond to ESA funding. If tuition increases are concentrated in schools at capacity, then accessibility once again does not really change.

It’s also important to note that the new, higher statewide average tuition amount in Iowa is still below the ESA amount! According to the paper’s data, average kindergarten tuition in 2023-24 was $6,800—six hundred dollars below the ESA amount. A student attending the average Iowa private primary school could still attend with zero out-of-pocket cost and have money left over for some books, tutoring, or a variety of other educational services.

2. TOO NARROW OF A SCOPE.

First, in the paper, the researchers state they have only been able to collect data on 51% of Iowa private schools, which serve 62% of Iowa’s private school students. Nebraska was no better—the researchers found data for 44% of Nebraska private schools, which account for 51% of Nebraska private school students. Considering there are around only 200 private schools in Iowa and Nebraska, respectively, it’s very unlikely enough schools are accounted for to be representative of the whole private school sectors in these states. This is especially true considering data availability is not random across schools, as the paper explains in its Data section.

Second, not enough data exists to establish parallel trends. The idea here is to know whether your treatment (in this case, a state with an ESA) and control (no ESA) were about the same in the category you’re interested in before the treatment was implemented.

In this case, we want to know whether Iowa’s average private school tuition was trending about the same way as Nebraska before Iowa started its ESA. If it was, then we can be more confident that a change in average tuition after the ESA launched is actually due to the ESA. This is especially important because the COVID-era ESSER payments have dramatically shifted the school finance landscape over the last several years. It’s important to know whether COVID affected private school tuition in both Iowa and Nebraska in the same way.

Unfortunately, because the paper only includes one year of pre-treatment data, we cannot confirm this. I assume this is something the researchers are working on. But education finance research studying post-COVID years is in dire need of solid parallel trends for its conclusions to be persuasive.

3. NEBRASKA IS AN INSUFFICIENT CONTROL.

To examine the effect of an ESA on Iowa, the researchers want to compare Iowa to an education environment that resembles Iowa but does not have an ESA. They have chosen Nebraska to fill this role because it will passed a universal ESA in 2023-24 and implement it in 2024-25.

There are a lot of problems here. First, Nebraska is not implementing a universal ESA in 2024-25. In the 2023 legislative session, Nebraska passed a tax-credit scholarship program (TCS). A TCS differs from an ESA in two ways. For one, funding does not come from public education expenditures. Instead, a tax credit is offered to individuals who donate to scholarship-granting organizations, who then take those donations and help students pay for private school tuition. Additionally, TCSs only pay for private school tuition, while ESAs can be used for a variety of educational expenses.

Second, Nebraska’s program is not universal. Most students can participate, but not all.

Third, Nebraska’s program future has faced lots of question marks. Immediately after its passage, some opposition led to a prospective ballot referendum in November 2024. In response, the 2024 legislative session saw Nebraska repeal the TCS and replace it with a voucher program. Now its funding mechanism is closer to a typical ESA, but the allowable expenses still differ.

Fourth, Nebraska has a budget cap on its voucher program. A maximum of 0.3% of the K-12 population of Nebraska could participate in the program.

If the researchers think a state with a not-yet-implemented ESA makes a great control state for an Iowa study, they must look elsewhere. But even Nebraska had a future ESA, it’d make a bad control state anyway. It’s very possible that Nebraska private schools may behave unusually if they know an ESA program will be implemented in another year. A school that typically slightly raises its tuition prices each year may decide to hold off on a tuition increase in 2023-24 if they know there will be an ESA in 2024-25. In that sense, a program is still impacting the decisions of Nebraska private schools.

Additionally, comparing only two states subjects the research to exogenous differences between the two states. Nebraska has less than half the population density of Iowa, even though it has the biggest metro area between the two states (Omaha). As such, the distribution of private schools looks very different in Iowa than it does in Nebraska, which means competitive effects between private schools will vary more than we would like for a study like this.

A better option would be collecting data across several states without large school choice programs on the books, as nothing ESA-related would be affecting private school decisions, and their various state differences compared to Iowa likely would average out. This requires much more work, of course, and may be part of the researchers’ long-term plan. For now, there’s not enough to make a difference-in-differences study work convincingly enough.

Understanding how ESA programs affect school tuition levels is an important question. It deserves a more rigorous approach than this paper offers.