Wrong answer on school finances
In the first week of October, the Massachusetts Supreme Judicial Court heard arguments in the Hancock school finance case. The arguments addressed the opinion issued last April by Superior Court Judge Margot Botsford that funding is constitutionally inadequate in the districts of those who brought the suit and, by extension, in high-poverty districts across the Commonwealth.
Much has changed since the SJC last examined school finance 10 years ago. Massachusetts now ranks fourth nationally in expenditures per pupil. The state has vastly reduced, eliminated, or even reversed spending gaps that previously existed between wealthy and poor districts, including the four plaintiff districts that are the focus of Hancock (Brockton, Lowell, Springfield, and Winchendon). Education Trust, a national organization devoted to narrowing achievement gaps, consistently finds that Massachusetts spends significantly more in poor districts than in wealthy ones, ranking at or near the top of the nation by various such measures.
In her opinion, Judge Botsford acknowledges this progress, but holds that funding is still inadequate. This opinion has been widely publicized, but the basis for it has not. A hard look at the evidence reveals a surprisingly weak case.
Adequacy studies inadequate
In order to demonstrate the inadequacy of funding, the plaintiffs commissioned two studies to estimate what an adequate education would cost in Massachusetts. The court found neither of these studies helpful, and it is important to understand why, for several reasons. First, it explains why the court’s opinion rests instead on what were clearly secondary arguments presented by the plaintiffs. It also explains why the court’s opinion that spending falls below what is necessary is unaccompanied by any indication of how much spending is necessary, since the court found no reliable basis for determining it. Finally, it is important to understand the shortcomings of these studies, since the nature of the opinion, if adopted by the SJC, may force the state into the very types of flawed studies that Judge Botsford rejected.
The plaintiffs’ two studies are representative of those produced by consultants for school finance cases, each of them following one of two frequently used methods: the “professional judgment” model and the”successful schools” model. The professional judgment approach asks educators to build an ideal school budget to meet certain educational objectives. In the plaintiffs’ study, those objectives were the seven state curriculum frameworks (English Language Arts; mathematics; science and technology; history and social science; foreign languages; the arts; and health). The panelists build the budget from the bottom up, by answering questions such as: What is the optimal class size? How many teacher aides and computers should there be? They are encouraged to “be creative and innovative,” to design new programs or services, and to assume there are no revenue constraints.
Predictably, the plaintiffs’ professional judgment study implied that almost every district in the state—even the wealthiest—was underfunded (with the ironic exception of Cambridge, a high-spending but low-performing district). Judge Botsford rejected this study as “a wish list of resources that teachers and administrators would like to have if they were creating an ideal school with no need to think about cost at all.”
In contrast, the “successful schools” approach is based on spending and performance data. The plaintiffs’ study selected the 75 districts with the highest MCAS scores on math and English Language Arts and went through a series of calculations to estimate what it would cost for other districts to attain the same level of success. For a variety of technical reasons, the calculations led to the nonsensical conclusion that two-thirds of the successful districts were not spending enough to be successful. Consequently, Judge Botsford also rejected this adequacy study.
Despite rejecting both of the plaintiffs’ attempts to calculate funding adequacy, Judge Botsford nonetheless concluded that the Commonwealth’s “foundation budget”—the state’s measure of the minimum cost for an adequate education—is too low to meet constitutional muster. But the basis for Judge Botsford’s conclusion is problematic.
Since the court discarded the plaintiffs’ direct estimates of the cost of adequacy, the court’s opinion relies instead on three strands of indirect evidence. The court relies most heavily on disparities in the ratio of spending to foundation budget between high-scoring and low-scoring districts. Second, the court’s opinion cites shortcomings in the foundation budget formula, especially for the cost of special education. Finally, the court sounds a note of alarm that “the funding inadequacies have been exacerbated by profound cuts in public school education funding by the Commonwealth in the last two years.”
The foundation budget is central to the funding-adequacy issue. As the state’s benchmark for the minimum cost of an adequate education, it is the cornerstone of state education financing. For any school district unable to fund schools at the foundation-budget level from property tax revenue, the state has pledged to make up the difference, and it has made good on that pledge even during the recent recession and fiscal crisis.
These premiums have provided a powerful mechanism for closing or reversing the spending gaps between poor and wealthy districts. But the foundation budget is the minimum every district must spend; it’s not a limit. Any school district is free to spend as much as localities choose, and many affluent communities fund their schools at levels well above the required minimum. As a result, while K-12 school districts in the poorest quartile spent, on average, more than $8,700 last year, districts with the least poverty spent about $8,500. Poor districts spent a bit more per-pupil than affluent districts, on average, but not by as big a difference as in foundation budget.
Here is what the plaintiffs made of that data. Rather than comparing per-pupil expenditure, they expressed spending as a percentage of foundation budget. By this measure, poor districts spend less than wealthy districts, even though, in real dollars, they spend more: Wealthy districts spend, on average, well above 100 percent of foundation budget, while average spending among poor districts is not much more than 100 percent. The plaintiffs postulate that wealthy districts would not spend more than necessary, so their foundation budget (about $6,700) must be inadequate for them to obtain their high level of performance. Thus, they conclude, the poor districts’ foundation budget (about $8,300) must be inadequate as well, even though it is far higher.
The court acknowledges the sketchy nature of this argument, describing the approach as offering a “rough” insight. Nonetheless, this is the main basis of Judge Botsford’s recommendation to the SJC regarding the constitutionality of our funding system.
Specifically, her opinion cites two facts as evidence that today’s foundation budget levels are inadequate: (1) the state’s highest scoring districts spend, on average, about 130 percent of foundation; (2) the state average is about 115 percent. From these facts, the court infers “it is difficult not to conclude that the minimum ‘adequate’ funding level for every school district in Massachusetts lies above the current foundation budget formula amount.”
This inference is not justified. Four flaws in the argument are as follows:
1. Average spending is no indication of necessary spending. The court uses the average ratio of spending to foundation budget to conclude that the “minimum adequate funding” ratio materially exceeds 100 percent for successful districts. However, among high-scoring districts, there is a wide spread of spending levels, including some that are close to the foundation budget minimum. Figure 1 plots MCAS scores (scaled on the state’s 100-point “proficiency index”) against percentage of foundation-budget spending for two groups of districts. The districts depicted by the green dots, which are in the lowest-poverty quartile, include almost all the highest-scoring districts, and their average spending is 125 percent of foundation. But the average is no indication of necessary spending, since a number of districts that obtain this high performance spend much closer to foundation. District M, for example, is at 104 percent, but its performance level is among the highest. More than one-fifth of the students who earn these high scores are in districts that spend less than 110 percent of foundation.
Similarly, it is wrong to infer anything about minimum necessary spending from the statewide average of 115 percent: So long as districts are free to spend more than the minimum, many of them will do so, and the average will exceed the minimum. To ignore this is to fall prey to the Lake Wobegon fallacy.
2. Association is not causation. On average, high-scoring districts spend more above foundation than low-scoring districts. From this fact, the court infers that higher spending is necessary for higher performance. But this confuses association with causation. After controlling for demographics, the association disappears.
Consider the black dots in Figure 1, representing districts in the highest poverty quartile (there are fewer of them, since they are larger districts). On average, these districts spend closer to foundation than low-poverty districts (106 percent vs. 125 percent). But these districts also score below low-poverty ones with the same or lower spending ratios. For example, each of the low-poverty districts spending 100 to 110 percent of foundation scores higher than all of the high-poverty districts, no matter how high they spend. Clearly, demographics are a powerful influence, so we cannot infer the effect of spending on performance without controlling for that.
To do so, look horizontally among black dots or green dots. There is no association between performance and percentage of foundation-budget spending. For example, among high-poverty districts, the worst performer, district H, spends nearly the highest, 119 percent. One of the best performers, district S, spends 123 percent of foundation, but district R does just as well at 101 percent. Turning to low-poverty districts, district M at 104 percent does about as well as district W at 158 percent. If we were to fit a line through the green dots or through the black dots, it would be virtually flat, signifying no association between performance and spending ratio among demographically similar districts.
Dozens of more formal analyses prepared for the court also show no such association. These analyses apply standard statistical methods to district and individual student data, using such demographic controls as district income and parental education, individual LEP, SPED, and free-and-reduced-lunch status. The court chose to disregard all such statistical analyses. In Judge Botsford’s opinion, the “real world” is better represented by simple averages, without attempting to control for demographics.
3. Comparing spending ratios of wealthy and poor districts uses a variable yardstick. Figure 2 depicts the same data as Figure 1 in terms of spending per pupil, instead of percentage of foundation. The minimum spending per pupil is higher in the high-poverty districts, because the foundation budget is much higher. After concluding, on a highly tenuous basis, that minimum spending is substantially too low in the wealthy districts, the court takes a further leap, concluding that minimum spending must be too low in the poor districts as well. As the diagram indicates, this rests on the assumed difference between minimum spending in poor and wealthy districts. In other words, the percent-of-foundation argument is that the base level of foundation (for students who are the least challenging to educate) is wrong but that the low-income premium is right.
The truth is, neither the base level nor the appropriate premium for low-income students can be known with certainty; at best, one can only determine a reasonable range, based on available evidence. That means policy-makers should have some latitude in designing a foundation formula. A progressive formula consists of a base level toward the low end of the range (since most wealthy districts will spend more anyway) and a low-income premium that is on the high side (since high-poverty districts are less likely to be able to exceed foundation spending levels). Massachusetts has a relatively high low-income premium, resulting in one of the nation’s most progressive funding systems. Reasonable proposals have been advanced to increase the low-income premium even more, most notably in Gov. Romney’s fiscal 2004 budget.
The irony is that raising the low-income premium (a measure recommended for consideration by the court) would not remedy the constitutional violation, under the court’s percentage-of-foundation logic. Raising foundation in poor districts from, say, $8,300 to $9,300 would increase spending by up to $1,000. But, as a percent of foundation, spending would be unchanged, still close to 100 percent. (The black dots in Figure 2 would move right, while those in Figure 1 would stay put.) For wealthy districts, with no change in either spending or foundation budget, the spending ratio would remain unchanged. Thus the disparity between average spending in relation to foundation budget in rich and poor districts would be undiminished. By the criterion used to justify the finding of inadequacy, it would be as if no solution had been adopted at all. Under the court’s criterion, no increase in the low-income premium, no matter how large, could eliminate the finding of inadequacy for low-income districts. This deterrent to a more progressive formula, no doubt unintended, vividly illustrates the perils of judicial intervention in policy-making.
4. There is no association between spending ratios and improvement over time. High test scores often reflect favorable demographics; whether scores are improving is often better evidence of what the school is actually doing. By focusing solely on a snapshot of test results, the court ignored evidence that a number of districts spending near foundation have shown strong improvement over time, while many districts spending well above foundation show little improvement. On average, high poverty districts have improved faster than low-poverty districts, despite lower spending ratios.
These four flaws show there are important differences in performance and rates of improvement among demographically similar districts that are unrelated to spending. There are even greater differences among schools within these poor districts. There are high-poverty schools whose scores on the proficiency index rival those of low-poverty schools and districts. How do we bring the lowest-performing districts (e.g. district H) up at least to the level achieved by the similarly poor district R, and to the higher levels yet achieved by some schools within these districts? That is the challenge in the next round of education reform, and it involves such elements as leadership, accountability, flexibility, and intervention, not funding levels.
Components of foundation
The second strand of the court’s argument is that certain components of the foundation budget formula underestimate true costs, most notably of special education. To evaluate this argument, one must consider changes in special-education funding made by the Legislature and the executive branch in recent years. The special-education component of foundation budget itself has been increased, and “circuit-breaker” legislation has been enacted to help pay for the extraordinary costs of severely disabled children. The circuit-breaker was implemented last year for the first time, providing an increase in funding over the program it replaced by 72 percent. As the court points out, this was still well below “full funding” of the new program, since claims jumped faster than expected. This year’s appropriation (enacted after the court’s report was issued) raises circuit-breaker funding by another 66 percent, to $201 million. State funding for special education cannot be evaluated on foundation formula alone without the circuit-breaker. And it would be premature to declare that circuit-breaker funding is inadequate, as its implementation is rapidly ratcheting up and the new program is still being evaluated.
The court also believes that the component of foundation budget for teacher salaries is inadequate. The basis for this conclusion is that the ratio of actual expenditures on salaries to that component of the foundation budget averages well over 100 percent and is particularly high in certain high-spending districts. But since teacher salaries are the largest component of the foundation budget, this simply reproduces the spending-to-foundation-ratio argument, with all its shortcomings. Moreover, it is misleading to decompose spending ratios and focus only on those components where spending exceeds foundation level, while ignoring those where districts often spend less than the foundation formula allots, such as support staff and central office.
Lessons from the downturn
The final strand of the court’s opinion involves spending cuts during the recent economic downturn, the result of what the judge called “profound cuts” in state aid. This strand is subordinate to the other two, since the court’s assertion here is that the cuts “exacerbated” funding inadequacies. But the data underlying this assertion are flawed. Specifically, the court reported in April that spending was expected to drop in each of the four plaintiff districts last year, most notably by over $10 million in Brockton. Instead, Brockton’s spending rose by about $2.6 million. The report also contains erroneous figures for fiscal year 2004 spending in the other plaintiff districts.
No dramatic decline in spending took place last year in any of the four plaintiff districts. Spending fell by about 1 percent in two of the districts, and rose by about 2 percent in the other two. Nor did spending drop as a percent of foundation, in three of the four districts. Finally, since enrollment dropped in each of these districts, spending per pupil fell only slightly in one district and rose by 2 or 3 percent in the others, in this last, hardest year of the fiscal crisis for public education.
More generally, the record of education spending during the three-year economic downturn is rather different from the impression conveyed by the court, as well as by press accounts and ads aired by the Massachusetts Teachers Association. From fiscal year 2001 to 2004, Massachusetts school spending grew by 12.7 percent (15 percent, if we include construction, transportation, grants, and federal funds). For the state as a whole, spending never fell in any year of the recession. On a per-pupil basis it grew 10.9 percent over this three-year period.
Spending growth certainly slowed during this period, but schools were generally able to avoid sharp revenue cuts due to the mix of local and state funding sources. For most districts, local tax revenues are the major source of school funds. Statewide, property taxes grow steadily at about 5 to 6 percent per year. That is because new development of residential and commercial property roughly doubles the base allowable tax growth of 2.5 percent under Proposition 2H. In general, local contributions to the schools closely track this steady growth in property taxes, in compliance with state law.
In contrast, state tax revenues are highly volatile. In fiscal 2002, state tax revenues fell by a devastating 14.6 percent. Education aid, however, based on a previously enacted budget, continued to rise at a healthy pace (including 15 percent hikes that year for three of the four Hancock districts). Over the next two years, as the fiscal crisis continued, state aid declined modestly. In all, by fiscal 2004 education aid remained 4 percent above fiscal 2001, despite a 4.5 percent drop in state revenues. The cuts in aid, though painful, were far more muted than in the last recession, when school aid fell 23 percent over three years, despite a much smaller drop in tax revenues.
For those districts with few local resources, the state was able to minimize cuts in aid. For the four Hancock districts, which rely on the state for 80 percent of their school funding, state aid rose 9.7 percent over this three-year period, about two-and-a-half times the state average. (The modest recovery in aid for fiscal 2005 also favors the Hancock districts. Their aid will grow 2.7 percent, or 4.2 percent per student, vs. 2.8 percent per student statewide.) This allowed their spending to grow 9.1 percent from 2001 to 2004. It would have grown 10.4 percent if the local contribution of these four districts had all kept pace with their tax levies, as it did in other low-income districts and in the rest of the state.
The lessons to be drawn from the downturn are not those drawn by the court, which seem to be that any cyclical slowdown of state education aid growth is a potential constitutional violation. Rather, one lesson is that because state revenues are far more volatile than local revenues, it would not be prudent for the Commonwealth to continually increase the share of education spending borne by the state. Aid should be carefully distributed to those communities whose local resources are most scarce, so that when the next downturn occurs, they can again be spared sharp cuts. Increasing the share of funds provided by the state could well make education spending more vulnerable to economic cycles.
An additional lesson from the downturn concerns collective bargaining. Figure 3 depicts the growth in education wages and employment in the four Hancock districts. (The picture, and the story it tells, is similar for the state as a whole.) Total education wages grew steadily throughout this period, from calendar year 2000 (the last year before the downturn) through 2003, for a rise of 14.8 percent in the four Hancock districts. (During the same period, total private sector wages in Massachusetts fell.) In the four Hancock districts, the average wage in education grew 13.4 percent during this period, exceeding inflation and several times higher than private sector average wage growth (3.3 percent). Education employment continued to grow in these districts through 2002, but layoffs occurred in 2003 as revenues became tighter while wages continued to rise unabated.
Several factors affect the growth of average wages, but the main factor is collective bargaining. In some districts, raises were rigidly locked-in prior to the downturn. Brockton, however, negotiated its teacher contract in January of 2003, well after the state’s revenues went into free fall. This was the month Gov. Romney came into office and determined that the state faced a $600 million deficit. He was authorized by the Legislature to implement immediate mid-year budget cuts, including local aid. And yet, Brockton negotiated a three-year contract raising school payroll costs by 11 to 12 percent. Under this contract, any individual with fewer than nine years of service could expect to receive raises of 25 to 37 percent over three years.
In response to testimony that districts such as Brockton made painful program cuts, the defense pointed out that those cuts wouldn’t have been necessary if they had negotiated less generous contracts. Judge Botsford dismissed this statement, and rejected the suggestion (that had not been made) that Brockton and Lowell teachers were “overpaid.” The court did not find that teachers in these districts were underpaid either. As the court reported, Brockton’s average teacher salary of $57,398 in 2001 exceeded that of Wellesley and Concord, two of the plaintiffs’ preferred “comparison districts.” The most recent data show that the average salary among the four Hancock districts exceeds the state average, which is seventh highest in the nation, according to the American Federation of Teachers. Nor did the court find persuasive evidence that salary levels kept plaintiff districts from attracting and retaining qualified teachers. In short, there was no evidence of any educational downside had the teachers’ unions showed modest restraint to avoid the painful program cuts cited by the court.
Repairing the foundation
The court’s opinion that funding is constitutionally inadequate is based on a weak case: None of the three strands of evidence stands up to scrutiny. Consequently, the court’s recommended remedy also rests on weak ground. That remedy would have the state revise its foundation budget within six months under court guidelines, and remain under judicial review, presumably until it meets the court’s criteria for adequacy.
The time is certainly ripe for the state to take a fresh look at the foundation budget, and it has already begun doing so. Unlike a decade ago, we now have considerable information on educational outputs to help move away from a strictly input-based approach. As a national leader in standards-based reform, Massachusetts is positioned to develop new methods to estimate the cost necessary to achieve continuing educational progress, as we consider raising the standard on math and English Language Arts and phasing in accountability for other subjects at a judicious rate. We can experiment with a foundation budget formula that guides districts toward the kind of reforms that are most promising. We need also to carefully consider if and how the foundation budget can be constructed to put educational needs over the sometimes-dysfunctional dynamics of collective bargaining.
None of this will happen if the state is handcuffed by a rigid and unrealistic court decree. Judge Botsford’s recommended court order would require the state to determine the cost of educating students to state standards in all seven curriculum frameworks. Our state standards still have a large aspirational component. Since only two of the frameworks (math and English Language Arts) are currently tested for graduation, there are no reliable data from which to estimate what it will take to generate acceptable student outcomes on the other subjects. Consequently, such an order would result in a highly speculative exercise, suffering from the same problem as all professional judgment studies: There is no attempt to tie observed spending levels to actual student outcomes.
After rejecting just such an estimate, produced for the plaintiffs, as a “wish list,” Judge Botsford would have the SJC order the state to produce another “wish list.” (Predictably, the Massachusetts Teachers Association has already filed a brief providing a host of such studies for emulation.) Should the SJC force the state into such inflated cost estimates, as a result of the state setting ambitious performance goals, there may well be a chilling effect on the standards movement, both here and elsewhere.
When considering evidence on individual elementary schools in Springfield, Judge Botsford found some high-poverty schools, spending less than their peers, that perform quite well—schools that outshine a number of the low-poverty districts depicted in Figure 1. The judge candidly notes “as a matter of puzzlement and concern…such great disparities in performance and quality. This variation within the same district, shown not to be particularly related to funding, indicates unaddressed problems of management and leadership.”At this stage of education reform, funding is neither the problem nor the solution. After committing billions of additional dollars to schools over the last decade, and leading the nation in establishing high standards and assessments, we have made great progress. State policy always envisioned that the next phase would focus on accountability and intervention to address the problems of management and leadership cited by the court. This will require resources to ramp up state efforts in educational auditing and intervention in low-performing schools and districts.
But it will also require a host of non-monetary measures, some as yet unknown, and some laid out in various reports, including that of the Governor’s Task Force on State Intervention in Under-Performing Districts, which is favorably cited by the court. Unfortunately, the teachers’ unions continue to block many of these measures. The thrust of the Hancock suit—funded almost entirely by the unions—is to redirect state policy away from key concerns flagged by school reformers, and back to another round of school funding hikes. If we become distracted by another huge monetary fix, our energies will be entirely absorbed by questions of how to finance it, how to distribute it, and so on, pushing aside the central issues of accountability, leadership, flexibility, and intervention that are so critical to fulfilling the promise of education reform.
Robert M. Costrell is chief economist for the Commonwealth, in the Executive Office for Administration and Finance, and professor of economics (on leave) at the University of Massachusetts in Amherst.