Tax Credit Policy Design for School Choice: A Response to John Kirtley

There are few people with whom I am so much in agreement on the goals of education policy as John Kirtley. To the extent that we differ, it’s chiefly about the best ways of achieving those shared goals. With that in mind, here are my thoughts on his recent post on Education RedefinEd.

Regulation of Private Schools under Education Tax Credit Programs

I’ve no reason to think that the customary financial reporting requirements imposed on scholarship-granting organizations (SGOs) are problematic, but the same cannot be said about regulations imposed on the private schools themselves. In response to an earlier post by Adam Schaeffer, John writes that “Adam is absolutely correct that you can only drive so much excellence through top-down accountability.” But Adam actually goes further, arguing that there is no evidence you can drive any excellence through “top-down accountability” (read: “government regulations”) on private schools. The purpose of noting the corruption in state schools (e.g., in Florida and Atlanta) is to show that even the vast array of government regulations imposed on public schools fails to curtail such defects. And, as I found in reviewing the worldwide literature comparing different types of government, pseudo-market, and market education systems, it is the least regulated, most market-like systems that consistently do the best job of serving families across all measured outcomes.

If there were compelling evidence that government regulations on schools could reduce corruption or boost academic achievement, there might be a case for such regulations being added to tax credit programs. But no such evidence exists to the best of my knowledge. In addition to failing to achieve its intended goals, regulation inhibits the educational freedom and diversity that are responsible for the market’s efficiency and responsiveness to families—undermining the whole purpose of a choice program.

The one and only empirically defensible argument in favor of regulations of schools under tax credit programs is political: in some states, it may not be possible to enact tax credit programs without such regulations. In that case, a judgment call has to be made on whether the regulations are so bad as to compromise the program and make it unworthy of passage, because it would fail to produce positive results and give the movement a black eye that would mistakenly be carried over to better, freer programs.

But that argument does not apply to states that have already enacted relatively free school choice programs, by definition: the bills have already passed, so the regs weren’t politically necessary.

Multi-SGO vs. Single SGO Tax Credit Program

 John also argues that a proliferation of SGOs increases the risk of a financial or other SGO scandal in any given year. It’s a plausible argument. If the probability of a scandal breaking out at any one SGO is fixed, and that probability is equal to x, then the probability of a scandal breaking out at any one of a group of SGOs would be = 1 – (1 – x)^N, where N is the number of SGOs. As N goes up, so does the likelihood of at least one scandal occurring.

On the other hand, it seems likely that the probability of a scandal breaking out at an SGO is not fixed, but rather is proportional to the number of people the SGO employs. Hiring more people raises the chance that you eventually hire someone crooked. Tempting as it is to develop a full-blown mathematical model for scandal risk based on the number and size of the SGOs, it seems sufficient to say that the two factors probably cancel each other out to a considerable extent. In other words, it is not obvious that scandal risk would be appreciably different in single SGO vs. multi-SGO environments.

But the absolutely crucial factor that the above discussion omits is that the impact of a scandal also varies with respect to the number of SGOs. If you have 200 SGOs, each of limited size, and one of them has a scandal, it’s no big deal. Donors will just stop giving to that SGO and it will go out of business unless it manages the Herculean task of convincing the public it has mended its ways. Parents that had gotten scholarships from it will then seek scholarships from one of the SGOs that is now receiving the donations that the corrupt one used to get. In fact, this is just what would have happened in the case of the errant donation that John mentions, whether or not any new regulations were imposed.

Indeed, we see this with charities in general. Charitable giving is hugely popular in the United States, but every year some charities are found to be fraudulent or mismanaged. This has had little or no effect on the popularity of charitable giving over time. The system has proven extremely resilient as former donors to the fallen charities have simply shifted their giving to better-run institutions. There has been no move to winnow down America’s charitable landscape either overtly or indirectly (such as by drastically limiting the share of donations that can be used for operational overhead, and thereby driving most institutions out of business).

John likens the probability of a scandal at an SGO in a multi-SGO tax credit program to “giving the enemy a hand grenade.” Extending that simile, the single SGO model is like giving opponents a nuclear time bomb. What happens if there is only one SGO that accepts all the donations in a state and it succumbs to a scandal? Donors have nowhere else to turn. Many will likely stop donating and poor families who were depending on those scholarships will suffer. That, in turn, will not sit well with legislators or the public, who would likely act to re-open the flow of funds to those families.

But how would they go about trying to do that on short notice? One way would be for the state to take over the corrupt SGO and promise to set it aright. As we’ve seen, state control of education does not prevent scandals, so that won’t work—but it would destroy the independence of the program. Another alternative would be for the state to create its own SGO alongside the corrupt one, with much the same effect. Yet another option would be for the state to impose a passel of new regulations on the corrupt SGO, promising that these would solve the problem—which, as we’ve seen, they’d be highly unlikely to do. And finally, the state could simply shut the program down entirely, forcing all those families back into the state school monopoly, which they had deliberately chosen to flee. Every one of those outcomes is far worse than the outcome in a state with many SGOs. Indeed, even without a scandal, a single-SGO system represents a vastly easier and more tempting target for a state takeover than a distributed, multi-SGO system.

I am both very happy and very relieved to grant that John’s organization, Step Up for Students, is impeccably run and is not likely to suffer a scandal in the foreseeable future. But, as Adam asks, what happens if/when John and the organization’s current top executives are gone? And similarly, what happens in states that don’t have a John Kirtley to create and brilliantly staff and oversee their one and only SGO? I’ve been around long enough to know that John Kirtleys do not grow on trees. The school choice movement is astonishingly lucky to have him as one of its leading lights. And because of the rarity of his invaluable qualities, it is unwise to design a policy that relies on similarly exceptional individuals to lead every SGO in every state in perpetuity.

Some SGOs will inevitably be overseen by less experienced, capable and honorable people (just as some public school districts are today) and so our education policies must systematically minimize the damage that mismanagement and corruption can do. Giving donors real choice among a panoply of different SGOs is the only mechanism yet suggested to accomplish this task, and one that has proven its value for generations in the broader charitable sector.

Preserving Parental Choice

There is another important reason to prefer the multi-SGO model, a goal that John and I deeply share: preserving real parental choice. A central conclusion of my journey through education history from classical Greece to modern America, England, Canada and Japan was that parents generally make better choices for their own children than even “expert” third parties make on their behalf. That is true whether the third parties are government bureaucracies or private institutions. And in my reading of that history I failed to find a single education system in which third parties who were paying for children’s education indefinitely restrained themselves from shaping what or how those children were taught. Ultimately, the money came with strings attached. (The G.I. Bill, sometimes presented as a counterfactual to this observation, is actually not relevant, since the pattern I’m describing is for elementary and secondary education. It is when children are young and their minds most malleable that the temptation is greatest to shape them in whatever image the third party wishes.)

We’d be wise to expect that pattern to continue. But if all third party payment ultimately comes with strings, how do you preserve the maximum level of parental choice? You prevent any single individual or organization from gaining a monopoly in third party education subsidies. By ensuring that a multiplicity of funding sources exists, you make it possible for parents to seek assistance from whichever organization most closely comports with their needs and preferences.

The freedom of donors to give to different SGOs that match their different educational ideals thus offers unique protection to families against being forced to accept strings they object to.

The alternative John suggests, legislatively forbidding SGOs from attaching any conditions to their scholarships, is unlikely to achieve its intended aim in the long run. Certainly there is no precedent for it, and it is easy to think of cases that would undermine it. What happens when schools begin to open that refuse to serve gay students? Or that exclusively serve them? What about Wiccan schools? What about “Marx’s Manifesto Middle School”? What about “Hayek High”? For each one of these schools, there is a distinct and sizable constituency that would deeply object to funding it. So what happens if all SGOs have to fund all of them? Do angry, coerced donor/taxpayers roll over and go gently into that good night of compulsion? No. They demand that their representatives impose restrictions on the eligibility criteria for private schools. But since we live in a pluralistic society, that ultimately will result in a gradual accretion of homogenizing restrictions on the kinds of education schools can offer, winnowing down the range of choices available to families—precisely the opposite of the intended goal.

Nor is this merely a hypothetical. I suspected this phenomenon would exist based on my research for Market Education: The Unknown History, but recently tested it statistically by comparing the level of regulation imposed on private schools under voucher versus education tax credit programs. Under vouchers, every taxpayer is compelled to pay for every government-approved private school. Under tax credit programs (with the partial exception of Florida’s) donors have very broad latitude in choosing the kind of SGO they will support, and SGOs have similar latitude. So, based on my theory, we would expect vouchers programs to result in more heavily regulated private schools than tax credit programs because of the compulsion that vouchers perpetuate but tax credits avoid. After crunching the numbers using two different statistical methods and allowing for thousands of different randomized ways of measuring regulatory burden, I found that vouchers do impose a large and statistically significant extra burden of regulation on private schools that tax credits do not. This evidence is not dispositive, but it is consistent with the analysis outlined above, and it’s the only evidence we have. I suggest that we ignore it not so much at our own peril as at the peril of the children we seek to serve.

One Model to Rule Them All?

In any discussion of the “best” policy, I think it’s wise to consider the possibility that any one of us (or even all of us) could be wrong. Education policy is hard. We can of course do our best to collect the widest possible body of relevant evidence, and to test our recommendations empirically whenever possible. We can and should have more discussions like this one, which are hugely valuable. But after all that, we could still be missing something.

Given that reality, there is one other step we can consider to maximize our chances of getting education policy right: we can let different models coexist side by side for a few decades and watch how well they perform, rather than trying to homogenize them up front. If one program doesn’t work quite as well as another, we will see that and be able to learn from it. But if we insist on conformity in the short term, we may never learn those important lessons. And if we settle on a fundamentally flawed model, the results could fall very far short of our hopes and expectations.

Of course, in any state that is just adopting a school choice program for the first time, there has to be a decision as to which program is best. But in states where programs have already been implemented, it seems far wiser to allow them to mature independently rather than to try to “fix” them or even supplant them with new and different programs simply because we believe that they may suffer shortcomings in the future.

Almost anything is better than the status quo monopoly, and yet we’ve had this awful monopoly for a century and a half. It’s hard to correct mistakes once a policy monoculture has established itself and crowded out the alternatives.