Earlier this week, NBER released the first random-assignment study ever to find a negative impact from a school voucher program. Previous gold standard studies had almost unanimously found modest positive effects from school choice, which raises the obvious question: what makes the Louisiana Scholarship Program (LSP) so different?

In an article for Education Next, I argued that, “although not conclusive, there is considerable evidence that the problem stemmed from poor program design.” The LSP is one of the most heavily regulated school choice programs in the nation, and that burden has led to a very low rate of private school participation.  Only about one-third of Louisiana private schools accept voucher students, a considerably lower rate than in most other states. From a survey of private school leaders conducted by Brian Kisida, Patrick J. Wolf, and Evan Rhinesmith for the American Enterprise Institute, we know that the primary reason private schools opted out of the voucher program was their concerns over the regulatory burden, particularly those regulations that threatened their character and identity. For example, voucher-accepting schools in Louisiana may not set their own admissions criteria, cannot charge families more than the value of the voucher (a meager $5,311 on average in 2012), and must administer the state test.

We also know from the NBER study that the participating and non-participating private schools differ in at least one important respect. Whereas the non-participating schools experienced modest growth over the decade before the voucher program was expanded statewide (about 3 percent, on average), the participating schools had been experiencing a significant decline in enrollment (about 13 percent, on average). In other words, schools that were able to attract students tended to reject the vouchers while voucher schools tended to be those where enrollment had been dropping.

The difference in enrollment trends suggests that the LSP’s regulatory burden had the opposite of its intended effect: discouraging higher-performing schools from participating, leaving only the lower-performing schools that were so desperate to reverse their declining enrollment and increase their funding that they were willing to do whatever the voucher program required.

Several other researchers and education reform advocates reached similar conclusions, including Matthew Ladner, Adam Peshek, Michael McShane, Lindsey Burke, and Jonathan Butcher. However, others expressed skepticism about what I shall call the Overregulation Theory, and proposed alternative explanations for the LSP’s poor results. 

Writing at Education Week, Douglas Harris of the Education Research Alliance for New Orleans concedes that “regulation probably does reduce the number of private schools, especially the number of higher-performing private schools,” but he still believes the Overregulation Theory is “premature.” Harris instead offers two potential alternatives: 1) the improved public/charter school performance in New Orleans made the performance of the private sector look relatively worse; and 2) the curriculum at most private schools may not have been aligned to the state test, so the poor performance merely reflects that lack of alignment rather than poor performance. 

Harris’s first theory is explicitly rejected by the NBER study. On the third page of the study, the authors write: “Negative voucher effects are not explained by the quality of public fallback options for LSP applicants: achievement levels at public schools attended by students lotteried out of the program are below the Louisiana average and comparable to scores in low- performing districts like New Orleans.” In other words, the public school alternatives are not so great and the performance of the participating private schools is considerably worse.

That said, Harris’s second theory, which Jason Richwine also suggested, is plausible as a contributing factor. However, it is no more plausible than the Overregulation Theory. Indeed, whereas the differences in enrollment trends between voucher and non-voucher private schools provide some suggestive evidence for the Overregulation Theory, Harris provides no evidence to support the Nonaligned Test Theory. How many voucher schools were already aligned with the state curriculum and/or administered the state test? At this point, we do not know. Moreover, to the extent that testing nonalignment explains some of the very large 0.4 standard deviation difference in math scores, it is unlikely that it explains all or even most of that difference. Then again, Harris stated that he will be releasing the results of his own research on the LSP, so it’s likely he knows something that I do not.

Harris also notes that the NBER study only examined the results of one year of one program. He is certainly correct that we need more data over time to draw firmer conclusions, which is one reason I presented my interpretation as “not conclusive” and wrote that “the regulations may have had the opposite of their intended effect” (emphasis added). And, indeed, there is some evidence that voucher schools improved slightly in the third year since the statewide expansion (although if the voucher schools were their own district, they’d still be the fifth-worst of 76 in the state).

Nevertheless, such strongly negative results should give reform advocates great pause about the regulatory strategies employed in Louisiana. We know the regulatory burden chased away most private schools, and we have evidence that the voucher-accepting schools had been struggling with declining enrollment. If we want to better understand the LSP’s atypically disastrous performance, its program design is the logical place to start.