Interstate Inequality in Educational Opportunity

Goodwin Liu?

For all that has been said about the nationalizing influence of the No Child Left Behind Act on education policy, one fact endures: States remain in the driver's seat on setting academic standards and distributing the resources needed to achieve results. After decades of state litigation and policy reform, there is some evidence that disparities in educational opportunity within states have lessened.[1] But a national goal of equal educational opportunity cannot be realized by addressing only inequality within states. The reason is simple: The most significant component of educational inequality nationally is not inequality within states but inequality between states. Even if intrastate disparities were eliminated, substantial disparities across states would remain. This fact casts a long shadow over the ideal of equal opportunity.

In this paper, I do four things. First, I describe current educational inequality across states in terms of funding, standards, and outcomes. Second, I show that interstate disparities in education resources have more to do with the capacity of states to finance education than with their willingness to do so, highlighting the need for a robust federal role in ameliorating interstate inequality. Third, I demonstrate how Title I reinforces rather than reduces interstate inequality in school funding. Fourth, I propose recommendations for reforming the federal role in school finance to be more responsive to state effort and capacity.

I. The Current State of Interstate Inequality

A. Education Spending

In recent decades, interstate inequality in school spending has been substantial and relatively constant in magnitude, with a north-south and east-west gradient reflecting the historical development of public education in the United States. Table 1 shows each state's per-pupil expenditure for 1969-70, 1979-80, 1989-90, and 1999-2000 in constant 1999-2000 dollars.[2] At the bottom of the table are two measures comparing the extent of interstate variation from year to year. The first is the ratio of the average per-pupil expenditure in the top ten states to the average per-pupil expenditure in the bottom ten states.[3] The second is the enrollment-weighted coefficient of variation, a measure of dispersion equal to the standard deviation as a percentage of the mean.[4]

Both the top quintile/bottom quintile ratio and the coefficient of variation show that interstate variation in per-pupil spending increased during the 1980s and then decreased during the 1990s. According to the coefficient of variation but not the ratio, interstate variation was somewhat less in 1999-2000 than in 1969-70. On both measures, the level of variation in 1999-2000 is comparable to the level that existed twenty years ago. While the extent of interstate variation has stayed fairly constant in recent decades, the relative standing of some states has changed significantly. In addition to per-pupil spending, Table 1 lists each state's rank for each year. The far right column shows the difference in rank for each state between 1969-70 and 1999-2000. On the whole, the national pattern of variation is fairly stable, with two-thirds of states moving no more than ten steps in either direction. But by increasing school funding at a rate significantly above the national average, a few states have moved up considerably in the ranking--for example, Georgia, whose economic growth has boosted education spending; Maine, where an increased state role in ensuring equity raised school spending in the 1980s; and Kentucky and West Virginia, whose legislatures overhauled their school finance systems after they were held unconstitutional.

Meanwhile, some states have moved down considerably as their per-pupil spending increased more slowly than other states'. The five states whose rankings fell the farthest--Arizona, California, Nevada, Utah, and Washington--are clustered in the West. In part, this reflects the political history of school finance reform, with California providing a familiar example. Yet robust increases in public school enrollment have also played a role. Arizona and Nevada, for example, saw the highest percentage increases in enrollment in the nation over the past three decades; each served more than twice the number of students in 1999-2000 than in 1969-70. Moreover, the relative decline in per-pupil spending in the West appears to be part of a broader trend. Among the twenty-six states whose ranking rose from 1969-70 to 1999-2000, only two--Texas and Wyoming--are located west of the Mississippi River. In sum, the map of educational inequality has become one in which the South, the Southwest, and far West trail the rest of the country.

The nominal spending data in Table 1 provide only a rough basis for interstate comparison for two reasons. First, there is considerable variation in the cost of providing the same educational services in different regions; for example, it costs more in New York than in Alabama to hire teachers of identical quality. To control for this, we need to apply a geographic cost index to equalize educational purchasing power across states.[5] Education economists have computed three leading cost indices, the most comprehensive of which is the Geographic Cost-of-Education Index (GCEI) developed by Jay Chambers.[6] This index estimates how much different jurisdictions must pay to hire a teacher with a given level of qualifications, taking into account the cost of living as well as key attributes of a region or school district that affect its attractiveness as a place to live and work. It then combines this model of teacher compensation with price indices for other school inputs to produce an index value for each state. Table 2 applies the GCEI to per-pupil spending data for 2001-02. Column A shows unadjusted per-pupil spending with state rank; Column B shows cost-adjusted figures.[7]

Second, states differ significantly in their student demographics and thus in the magnitude of their educational task. Although North Dakota and Texas have comparable per-pupil spending, for example, Texas faces a greater educational challenge because a higher percentage of its children are poor or LEP. In order to meaningfully compare spending across states, we need to know "the extent to which [states] with a harsh educational environment, as measured by the characteristics of their students, must pay more to achieve the same performance as other [states]."[8] Ideally we would estimate educational resource needs at an individual level based on each student's family background, school and neighborhood environment, past academic achievement, and other factors. But because such data are not available on a national basis, adjustments for student need are typically done by weighting enrollment data based on the number of students belonging to groups known to require additional resources to attain a given performance level. To adjust per-pupil spending for student needs, I assigned a weight of 1.6 to students from poor families (in other words, poor students are estimated to require 60% more resources than non-poor students),[9] 1.9 to students with disabilities,[10] and 1.2 to LEP students.[11] I then divided each state's total cost-adjusted expenditures by its weighted pupil count to derive its cost-adjusted spending per weighted pupil.[12] Column C of Table 2 lists these results in rank order.

As Table 2 shows, adjusting for cost and student needs reduces overall variation across states, but the extent of variation remains substantial. The top ten states in Column C spent an average of $7861 per weighted pupil in 2001-02, which was nearly 50% more than the $5292 per weighted pupil spent by the bottom ten states. While the cost of providing education tends to be lower in low-spending states, such states tend to have higher percentages of students with special needs. West Virginia, ranked tenth in Column B, drops to seventeenth in Column C largely because its child poverty rate is over 20%, compared to 15% nationally. New Mexico, ranked thirty-second in Column B, drops to fortieth in Column C; 24% of its children are poor, and 20% are LEP.

Tables 3a and 3b compare the demographics of students in high- and low-spending states. Whereas the student body in the top third of states is 70% white, 12% poor, and 4% LEP, the student body in the bottom third is 50% white, 17% poor, and 13% LEP. Black students appear evenly distributed across high- and low-spending states. But the states in the bottom third of spending, while enrolling 47% of the nation's schoolchildren, serve 54% of all poor students, 75% of all Latino students, and 76% of all LEP students. By contrast, the states in the top third enroll 29% of all schoolchildren, but only 24% of the nation's poor students, 16% of Latino students, and 13% of LEP students. In short, children with the greatest educational needs live disproportionately in states with the lowest education spending. As Column C of Table 2 shows, the bottom third is exclusively comprised of states in the South, Southwest, and West.

We can better comprehend the magnitude of interstate spending disparities by comparing them to intrastate disparities. I obtained data from the National Center for Education Statistics on the per-pupil expenditure of unified school districts at the 10th, 50th, and 90th percentile of spending in each state in 2001-02.[13] These data, adjusted for differences in educational costs and student needs, appear in Table 4. What we observe is that large intrastate disparities exist in jurisdictions like Colorado, New York, and North Dakota, while disparities are much smaller in states like Alabama, Kentucky, and West Virginia. Intrastate disparity is positively correlated with median district spending; states with higher spending tend to have greater interdistrict disparity. High-spending states with a large expenditure range tend to be comprised of numerous small school districts, whereas low-spending states with a small expenditure range tend to be dominated by large countywide school districts.[14] For all states, the range of variation below the median is smaller than the range above the median.

Figure 1a uses these data to illustrate the large interdistrict disparities across states. For each state, the bar represents the range of expenditures from the 10th percentile to the median. As the figure shows, the 10th percentile districts in fourteen states (Wyoming to Kansas) spend more than the median districts in fifteen states (Louisiana to Arizona). In other words, even if school finance reform in the fifteen low-spending states were to raise spending in the bottom half of districts up to the state median, those districts would still trail 90% of districts in the fourteen high-spending states. Similarly, Figure 1b shows that the median districts in eleven high-spending states (Alaska to Maine) spend more than the 90th percentile districts in eleven low-spending states (North Carolina to Florida). Finally, Figure 1c depicts the starkest interstate inequalities. The 10th percentile districts in eight high-spending states (Wyoming to Delaware) have per-pupil spending within $500 of the amount spent by the 90th percentile district in eight low-spending states (California to Florida). Consistent with these data, other studies report that interstate disparities account for well over half of the total extent of interdistrict inequality throughout the nation.[15]

B. Educational Standards and Outcomes

Since 1990, the National Assessment of Educational Progress (NAEP) has provided a valid basis for comparing student achievement across states. With data from state NAEP tests and from each state's own assessment system, we can observe variation in educational standards and outcomes across states.

Figures 2a and 2b compare the percentage of fourth-graders in each state achieving a "proficient" score on 2005 NAEP math and reading tests with the percentage of fourth-graders achieving a "proficient" score on 2005 state tests.[16] In each graph, the solid sloping line shows where states would line up if their proficiency standards matched NAEP's. The dotted sloping line is the best-fit line indicating the relationship between NAEP and state tests in an "average" state. The vertical line marks the percentage of students nationally who scored proficient on NAEP. From these graphs, we learn three things.

First, state standards of academic proficiency are literally all over the map and are mostly less rigorous than NAEP's. In Tennessee, for example, 87% of fourth-graders achieved a proficient score on the state math test, but only 28% scored proficient on NAEP. Similarly, 83% of students in Alabama were proficient on the state reading test while only 22% were proficient on NAEP. By contrast, states like Maine, Massachusetts, South Carolina, and Wyoming have proficiency standards that approximate NAEP's. This wide-ranging patchwork of educational standards is unsurprising in view of the broad discretion states have to define what content their students should know, how well they should know it, and what assessments are used to hold schools accountable.

Second, student performance varies considerably from state to state when measured against a common standard. While 35% of fourth-graders nationwide achieved proficiency on the NAEP math test, state figures ranged from 49% in Massachusetts and 47% in Kansas and Minnesota to 21% in Alabama and 19% in Mississippi and New Mexico. Likewise, the share of students scoring proficient on the NAEP reading test varied from 44% in Massachusetts and 38% in Connecticut and Minnesota to 20% in Louisiana and New Mexico and 18% in Mississippi, with 30% proficient nationwide. NAEP also reports scores in math and reading for all grade levels on a single 500-point scale. Those data show that the average fourth-grader in Massachusetts, Minnesota, and Vermont scored almost twenty points higher in math and reading than her peers in Alabama, Mississippi, and New Mexico--a difference of roughly two grade levels.[17]

Third, the states with NAEP proficiency rates lower than the national average are almost all low-spending states in the South, Southwest, and far West. Among the twenty-one states to the left of the vertical line in either Figure 2a or Figure 2b, only three (Georgia, Oregon, and West Virginia) are in the top half of the nation in terms of adjusted per-pupil spending. Conversely, while a few low-spending states have above-average rates of proficiency on NAEP in math and reading (e.g., Idaho, South Dakota, and Washington), the vast majority of high-performing states are high-spending.

Although this pattern suggests a relationship between resources and outcomes, it is important to remember that low-spending states have a disproportionate share of poor, minority, and LEP children. Student demographics, parental education and income, and other aspects of family background undoubtedly play a role in explaining performance disparities across states. Moreover, states vary in how they spend education funds, in their degree of intrastate finance equity, in the standards they set for teachers and students, and in the policy and regulatory environment they establish for schools and districts. All of these factors complicate the relationship between resources and results.[18]

But the notion that students in low-spending states would benefit from additional resources need not depend on a clean linear relationship between dollars and achievement gains. One might expect the relationship to be stronger where current spending is low and somewhat weaker or unpredictable where spending is already high. This intuition is a reasonable inference from the principle of marginal utility, which predicts that additional resources will make the greatest difference to those who have the least. As it turns out, this view is supported by the leading empirical study of state NAEP results, published by RAND in 2000.[19]

Using NAEP math and reading scores from forty-four states between 1990 and 1996, the RAND study compared performance across states to determine the efficacy of varying levels of per-pupil spending and varying approaches to resource utilization. Controlling for parental education, income, race, family size, single-parent status, and other socioeconomic status (SES) indicators, the study found that variation in state NAEP scores fell within a range of one-third of a standard deviation on a national scale.[20] In other words, students in the highest-scoring states were roughly one and one-third grade levels ahead of similar students in the lowest-scoring states.[21] Some low-spending states (e.g., Texas, Missouri) performed better than the average state, and some high-spending states (e.g., Rhode Island, Vermont) performed worse. But overall, spending was positively correlated with performance when similar students were compared.[22]

The study went on to investigate what uses of resources were most effective. The authors found that increased performance on NAEP was associated with additional resources for increasing participation in public prekindergarten (pre-K) programs, for lowering pupil-teacher ratios in grades one to four, and for improving instructional materials and resources for teachers.[23] Moreover--and this is a key finding--the size of the effect of lowering pupil-teacher ratios in early grades varied inversely with family socioeconomic status: children from low-SES families gained more from lower pupil-teacher ratios than children from medium-SES families, and the latter gained more than children from high-SES families.[24] The study similarly found that children from low-SES families benefited more from greater access to public pre-K programs than children from medium-SES families, who in turn benefited more than children from high-SES families.[25]

These findings suggest that resource-dependent interventions are most effective when targeted to low-SES states and, within states, to low-SES districts and schools.[26] Although the RAND study has its skeptics,[27] its results cohere with three other lines of empirical study that find positive resource effects on the performance of the most disadvantaged students and schools. First, randomized experiments on class size reduction--notable for their rigorous research design[28]--have found that smaller classes produce gains by all students but significantly larger gains by minority students, low-income students, and low-achieving students compared to their more advantaged peers.[29] Second, some econometric studies have similarly found that greater resources are associated with greater gains by low-achieving students relative to their high-achieving peers and by students in low-spending versus high-spending districts.[30] Third, from the late 1960s to early 1990s, increased education spending largely in the form of compensatory programs for low-income children coincided with robust gains in reading and math by black, Latino, and low-scoring white students, with the greatest gains in the South, even as the broad majority of whites made little or no improvement.[31] Changes in parental income and education explain only part of the gains by disadvantaged students,[32] and investments in schooling over this period, including substantial reductions in pupil-teacher ratios, had differential positive effects for disadvantaged students.[33] In sum, this body of evidence supports the common-sense inference that additional resources are likely to produce educational benefits--indeed, the greatest benefits--for the disadvantaged children concentrated in the lowest-spending states.

potential ability to raise revenue from its own sources."[34] In other words, fiscal capacity is an inherent characteristic of a state's economy and revenue base rather than a function of its decisions about how to raise revenue.[35] So defined, fiscal capacity can be measured in various imperfect ways.[36] Although state personal income (SPI) and gross state product (GSP) are two common measures, here I choose a more comprehensive measure of state fiscal capacity called Total Taxable Resources (TTR).[37] Introduced in 1985 by the Treasury Department, TTR is estimated by taking GSP as a starting point, subtracting payments to the federal government that states cannot legally tax, and then adding several income flows, including resident wages from out-of-state employment, dividends and interest income, and payments from federal social insurance programs.[38] In recent studies by the U.S. Government Accountability Office (GAO), TTR has been GAO's preferred measure of state capacity to fund public services, including education.[39]

We can compare capacity to finance education across states by computing each state's cost-adjusted TTR per weighted pupil.[40] Column A of Table 5 lists these data for 2001 in rank order, along with each state's ratio to the national average. As Column A shows, there are substantial differences in state fiscal capacity. Most states in the Northeast and upper Midwest are above the national average, while most states in the South and Southwest are below average. The fiscal capacity of the top quintile of states taken as a whole ($238,000 per weighted pupil) is over 57% greater than the capacity of the bottom quintile ($151,000 per weighted pupil).

Turning now to effort, each state's educational effort may be defined as the hypothetical tax rate that, when levied against the state's fiscal capacity, produces the observed level of nonfederal education revenue in that state. The tax rate is hypothetical because no such tax is actually levied; in almost all states, nonfederal education revenue is derived from a combination of state and local sources at various tax rates. At the same time, the definition assumes that the level of nonfederal education revenue in a state is a function of policy choices within the state's control. Thus, effort is an aggregate measure of the state's willingness to leverage available resources for education.[41]

To measure effort, I begin with each state's cost-adjusted revenue per weighted pupil from nonfederal sources in 2001-02; these appear in Column C of Table 5 along with ratios to the national average.[42] With these data, each state's educational effort can be determined by taking nonfederal revenue per weighted pupil as a percentage of state fiscal capacity per weighted pupil. The results appear in Column B, along with ratios to the national average. Like fiscal capacity, effort varies across states. However, a regional pattern is difficult to discern.

Table 5 provides some insights into the nature of school funding disparities across states. Some states, like New Jersey and New York, combine high fiscal capacity with above-average effort to generate a much higher level of education revenue than in most other states. Other states, like Maryland and Massachusetts, can achieve high revenue with below-average effort because of their high fiscal capacities. Delaware, home to many corporate headquarters, exerts the lowest level of effort but still has high revenue per pupil because it has the highest fiscal capacity in the nation. By contrast, some states generate high revenue (e.g., Maine, Michigan) or average revenue (e.g., South Carolina, West Virginia) by exerting high effort against low fiscal capacities. Among states with low revenue, many exert average effort (e.g., Arizona, Oklahoma) or even above-average effort (e.g., Arkansas, New Mexico) but draw limited revenue because of low capacity. Other states have low capacity and low effort (e.g., California, Louisiana), while some appear to have low revenue primarily because of low effort (e.g., Florida, Nevada).

These examples show that both effort and capacity play a role in explaining interstate disparities in educational resources. We can gauge the relative importance of the two factors by comparing the relationship between capacity and revenue with the relationship between effort and revenue. Table 6 describes these relationships with simple correlation coefficients using TTR, SPI, and GSP as alternative measures of fiscal capacity. Using unadjusted data on revenue and capacity, we find that, while revenue is positively associated with both capacity and effort (top panel), the relationship between revenue and capacity is much stronger. When the data are adjusted for geographic cost differences and pupil weights (bottom panel), there is an attenuated but similar difference between capacity and effort as a correlate of state revenue. Thus, while some states with low capacity manage to achieve high revenue with high effort, and while others with high capacity have low revenue because of low effort, Table 6 suggests that variation in fiscal capacity plays a larger role in explaining interstate differences in nonfederal education revenue than variation in effort.

The advantage of high fiscal capacity is further evident from the negative correlation between state capacity and state effort.[43] In other words, states with higher capacity tend to exert lower effort. Among the ten states with the highest fiscal capacity, only two exerted above-average effort in 2001-02, and neither one exceeded the average by more than 10%. By contrast, among the ten states with the lowest capacity, eight showed above-average effort, and four exceeded the average by more than 10%. Despite the generally higher effort exerted by states with lower capacity, nonfederal revenue per weighted pupil was almost 40% greater on average in the ten states with the highest capacity ($7615) than in the ten states with the lowest capacity ($5480). This pattern is analogous to the familiar inequality between school districts in states that rely heavily on local property taxes to fund education.

In sum, fiscal capacity and effort are both determinants of interstate disparities in educational resources, and between the two, capacity plays the larger role. States with higher capacity tend to make less effort yet raise more revenue than states with lower capacity. This reality highlights the need for a robust federal role in ameliorating interstate inequality.

III. The Federal Role in Interstate Inequality

Yet the federal government has done little to narrow educational inequality across states. The federal role in education, while greatly expanded by the No Child Left Behind Act, does not set common content or performance standards for schools in every state.[44] Nor does it seriously address interstate inequality in school funding. On the whole, federal spending on public elementary and secondary schools is small, comprising 7.9% of total education revenue in 2001-02.[45] Although federal aid disproportionately benefits poorer states, the equalizing effect is modest. Counting only state and local revenue, cost-adjusted revenue per weighted pupil in 2001-02 was 50% greater in the ten highest states as a whole ($8180) than in the ten lowest states ($5438). Taking federal revenue into account, cost-adjusted revenue per weighted pupil remained 44% greater in the ten highest states ($8,745) than in the ten lowest ($6056). The addition of federal funds to state and local revenue reduced the coefficient of interstate variation in cost-adjusted revenue per weighted pupil by only 11%.[46] In short, the federal government cannot buy much equality with eight cents of every education dollar.

The limited leverage of the federal share is a function not only of its small size but also of the way it is allocated. Federal education aid largely flows through categorical programs, not through general assistance grants. Among the three biggest programs, two--special education for children with disabilities and nutritional aid for low-income children--allocate funds largely in proportion to each state's share of the target population. These monies account for the mildly equalizing effect of federal aid across states because low-spending states tend to have higher shares of low-income children and because equal federal dollars per child provide a bigger boost, proportionally speaking, to low-spending states than to high-spending states. However, the single largest federal investment in the nation's public schools, Title I, does not reduce but instead reinforces interstate inequality in educational opportunity.

With over $13 billion appropriated in 2005, Title I aims to ensure equal educational opportunity for all children throughout the nation, whether poor, minority, or limited in English proficiency. Given this broad ambition, one might expect Title I to disproportionately benefit low-spending states, where disadvantaged students are concentrated. But the reality is otherwise. Like the bulk of equity-based policy and litigation in recent decades, Title I primarily works to reduce educational inequality within states, not between states.

The reason is simple. Each state's Title I allocation is largely a product of two factors. The first factor--the number and concentration of poor children in the school districts of each state[47]--tends to benefit low-spending states because they have disproportionate numbers of poor children. However, the second factor--"the average per-pupil expenditure in the State" (the state expenditure factor)[48]--causes the existing pattern of interstate inequality in education spending to be reproduced in the allocation of Title I funds. Although the statute limits the state expenditure factor to a range from 80% to 120% of the national average,[49] significant interstate disparities remain.

These disparities are evident in Table 7. Column A lists the number and percentage of the nation's poor children in each state in 2001, and Column B lists each state's share of Title I funds in 2001.[50] Together, Columns A and B show that high- and low-spending states do not receive Title I money in proportion to their shares of the nation's poor children. Michigan, for example, had slightly more poor children than North Carolina but received well over twice as much Title I aid. Similarly, Massachusetts had fewer poor children than Oklahoma but received almost 80% more Title I aid. Column C shows each state's Title I funding per poor child in rank order. Some of the highest amounts in Column C reflect statutorily guaranteed minimum allocations for small states.[51] Leaving those states aside, the amounts per poor child at the top are as much as double the amounts at the bottom, with the variation essentially mirroring interstate variation in per-pupil spending.

Of course, by channeling aid to high-poverty districts, Title I has the effect of narrowing disparities in educational opportunity for poor versus non-poor children. Federal education aid is significantly more targeted to poor children than either state or local funding.[52] However, as Table 7 suggests, the equalizing effect occurs only within states, not across states, because of the state expenditure factor in the Title I formula.

The disparities in Table 7 are somewhat overstated because the dollar figures are not adjusted for geographic cost differences. But even when cost adjustments are applied, the state expenditure factor effectively neutralizes whatever interstate equalization Title I achieves as a result of targeting funds to poor children. Indeed, the addition of Title I funds leaves the extent of interstate variation in revenue per weighted pupil virtually unchanged.[53] What the poverty factor in Title I does for interstate equalization, the state expenditure factor negates. Thus, remarkably, the mildly equalizing effect that the totality of federal education aid has across states occurs not because of, but in spite of Title I.

What is especially troubling is that this distribution of federal aid serves no convincing policy rationale.[54] The state expenditure factor cannot be said to adjust Title I allocations for geographic differences in educational costs, since state expenditures vary for many reasons having nothing to do with interstate cost differences. Even on a cost-adjusted basis, Title I allocations per poor child vary substantially across states. Nor can Title I be said to reward state effort; as discussed above, state per-pupil expenditure is more closely associated with state fiscal capacity than with state effort.

Moreover, the Title I formula cannot be understood to create an incentive for states and school districts to devote more of their own resources to public education. Title I aid is simply too small for this purpose. Suppose, for example, that Mississippi in 2000-01 had raised its per-pupil spending by $100 from $5175 to $5275, a 1.9% increase.[55] Assuming that Title I aid increases proportionally, Mississippi would have received $160 million under Title I in 2003 instead of $157 million, an increase of $3 million.[56] However, this increment is just six percent of the $50 million that Mississippi would have had to spend to raise its per-pupil average by $100.[57] As Congress's own researchers have observed, "[i]t seems unlikely that such a relatively small 'bonus' would provide substantial motivation to states and [school districts] in deciding whether to increase their level of spending for public elementary and secondary education."[58]

A further possible rationale for the state expenditure factor is largely historical. Four decades ago, when Title I was enacted, the weak condition of public education throughout the nation was evident not only in low per-pupil spending but also in feeble infrastructure at the state level. The Senate report on the Elementary and Secondary Education Act cited the example of "a medium-sized department in a middle-income State" where "75 professional staff members assist 1,300 schools and 20,000 local school people in the administration of State and Federal funds and programs . . . but these 75 State consultants can visit the schools of their State on the average of only one-half day every 7 years."[59] In this context, calibrating Title I aid to state expenditures might have ensured that states did not receive more funds than they could use efficiently. In 1965, Title I had the effect of significantly increasing the education budget of some states; in some schools, the new program increased funding by as much as 50%.[60] The ability of states and their subunits to effectively utilize this infusion of resources was not yet known, and the early years of Title I saw some instances of malfeasance.[61]

Forty years later, the educational infrastructure in most if not all states has become stronger. Their capacity to plan, implement, and evaluate educational programs has grown, as control of policy and funding has drifted upward from local school boards to large and professionalized state departments of education. Equally important, Title I comprises a smaller share of education budgets today than forty years ago. As a result, Title I's marginal impact on state administrative capacity is much less now than it was in 1965. Moreover, the current statute authorizes states to devote a portion of Title I money to administration, evaluation, and technical assistance in order to enhance the efficacy of program funds.[62] These considerations tend to erode any justification for the state expenditure factor as a means of limiting Title I grants to what states can effectively use.

Nor is it convincing to suggest that the state expenditure factor reflects a policy of deference to diversity in educational approaches among the states. Of course, there is no single, optimal level of per-pupil spending given the many combinations of resources, accountability, choice, and other variables that potentially comprise an effective state education policy. At the margin, it may be unclear what difference an additional hundred dollars per pupil will make in a given state, and Congress may reasonably wish to encourage variation. But as Table 7 shows, the disparities in Title I allocations are not marginal but quite substantial. It is perverse to justify this scheme as a kind of national experiment to test whether low-spending states can educate poor children equally well with one-half or two-thirds of the resources available in high-spending states. Such inequality may spur innovation, but only with unacceptable risks. To my knowledge, the state expenditure factor has never been defended in these terms.

Reforming the Federal Role in School Finance

Just as a patchwork of state standards offers little guidance for educating a national citizenry, a patchwork of state funding practices reflecting disparate levels of fiscal capacity and effort cannot effectively support ambitious national education goals. Narrowing those disparities ought to be a central focus of the federal role in school finance.

Second, in aiding states with low education spending, federal policy should distinguish between low fiscal capacity and low effort. Where low spending is due to low effort, the primary federal role should be to motivate states toward greater effort. Similarly, the federal government should ensure that states receiving increased federal aid do not reduce their effort or use federal money to supplant state or local funds. The reality is that, even with an expanded federal role, states will continue to bear most of the burden for school finance. Because a fully federalized finance system is neither realistic nor desirable, narrowing interstate disparities will require a progressive distribution of federal aid that is layered on top of a commitment by each state to do its fair share.

Third, federal aid should take into account geographic differences in educational costs. Because educational purchasing power varies significantly between states and within states, the efficacy of federal aid in reducing real differences in opportunity requires that cost differences be part of the equation.

Finally, federal aid will do much to reduce interstate disparities or motivate states to adopt high standards so long as it is only eight cents of every education dollar. Because the federal government has assumed a leading role in standards-based reform, it is fair to expect increased federal responsibility for the associated costs. Indeed, there is growing evidence that the ambitions of standards-based reform demand significantly more resources than what is now being committed.[63] Although Title I was once limited to remedial instruction for poor children, today it drives a systemic national agenda of standards and accountability. As Allan Odden and Lori Kim have observed, "some type of nationwide base per-pupil spending level is the logical school finance policy for the implementation of national education goals, especially since spending differs across states and spending differences are correlated with a variety of student outcomes."[64]

In sum, the federal role in school finance, in addition to targeting aid to the neediest districts and schools in each state, should (a) promote interstate equality by compensating for interstate disparities in fiscal capacity, (b) motivate states to exert reasonable effort in support of education, (c) adjust federal aid for geographic cost differences, and (d) provide foundation aid that is sufficient to enable even the poorest states to educate their children to national standards.

B. Policy Recommendations

With these principles in mind, I offer two recommendations for reshaping federal education aid. One is a modest proposal to reform Title I. The other is a more ambitious proposal to subsume Title I within a larger national program of foundation aid that would guarantee each state, whatever its fiscal capacity, a minimum level of educational resources per weighted pupil. A national foundation program would not achieve absolute equality, since states may always spend above the foundation level. But it would create a more equitable system of school finance and one that guarantees every child an adequate opportunity for equal citizenship in the national community.[65]

1. Reforming Title I. The state expenditure factor in the Title I formula should be eliminated. This reform would bring Title I into line with the aid formulas for special education, English language instruction, and child nutrition, all of which assign equal weight to eligible children regardless of the state where they reside. Title I should simply allocate aid in proportion to each state's share of poor children and apply a cost factor to adjust for geographic cost differences.

Although this reform would make Title I more equitable, its impact on interstate inequality would be modest because Title I would continue to provide only a thin layer of federal categorical aid on top of large interstate disparities in nonfederal education revenue. Any serious effort to reduce interstate inequality must directly address the wide variation in state effort and fiscal capacity. This can be done through a national program of foundation aid that complements the systemic reach of NCLB and the plausible evolution of federal policy toward national standards.

2. Creating a national foundation plan. There are many ways to design a foundation program that compensates for interstate disparities in fiscal capacity. One approach is a modified form of "power equalizing" whereby the federal government would guarantee each state a minimum amount per weighted pupil for a given level of state effort.[66] For example, the government could assure each state an amount per weighted pupil at least equal to what the state would have raised had it applied its tax effort against the average fiscal capacity among all states. For poorer states, whose actual revenue at a given level of effort is less than the guaranteed amount, federal aid would make up the difference. Richer states whose actual revenue exceeds the guaranteed amount would retain their revenue but would receive no aid. Under this scheme, federal aid would boost the fiscal capacity of poorer states while leaving wealthier states to their superior means, thereby narrowing (though not eliminating) interstate inequality. Moreover, by treating weighted pupils as the unit of analysis, the funding scheme integrates the compensatory thrust of categorical aid like Title I.

This type of program is a step in the right direction, although three modifications are warranted. First, if an important objective is to establish a national foundation of aid, then the program must specify a minimum level of effort that participating states must meet. The foundation program should not function as insurance against state indifference. Instead, it should serve as a framework for cooperation federalism in which the federal government would guarantee to every state exerting the minimum effort a foundation level of spending per weighted pupil.[67] Although a state conceivably could refuse to make the required minimum effort, any serious program of national foundation aid would involve large sums of federal money that states would find difficult to forgo.

Second, although it would be equitable to limit federal aid to low-capacity states, a power-equalizing foundation program is unlikely to succeed politically unless it spreads federal aid widely so that every state receives some. Instead of offering no aid to wealthier states that already exceed the federally guaranteed amount at any given effort level, a better approach would be a graduated system that provides some aid to every state. One example of this approach is the variable "federal medical assistance percentage" used by Medicaid. Under Medicaid, the federal government matches state spending on health-related services for low-income people at a rate that is different for each state depending on the square of the ratio of its per capita income to national per capita income.[68] States with lower per capita income have a higher federal matching rate, and states with higher per capita income have a lower matching rate, with all rates bounded by a minimum of 50% and a maximum of 83%.[69]

An analogous "federal educational assistance percentage" could be created to provide foundation aid to public schools. For each state at or above a minimum effort level, the federal government would match its cost-adjusted education spending per weighted pupil at a rate that takes into account the state's fiscal capacity relative to the average fiscal capacity among all states. Fiscal capacity would be measured by a state's total taxable resources adjusted for geographic cost differences and then divided by its weighted pupil count. For poorer states, the federal matching rate would be higher and, for the poorest states, high enough to ensure an educationally adequate foundation. For wealthier states, the matching rate would be lower and, for the wealthiest states, bounded by a politically acceptable minimum (say, four percent).

Third, the federal aid program will not serve its purpose unless it furthers not only interstate but also intrastate equality. If we wish to ensure a foundation level of resources per weighted pupil, it makes little sense to allow states to channel large portions of federal aid toward the most advantaged districts or the most advantaged students. To participate in the program, each state should be required to use federal aid not only to bring all districts up to at least the foundation level[70] but also to narrow both interdistrict and intradistrict resource disparities.[71] One approach would be to require each state to use federal aid to reduce its coefficient of interdistrict variation by a minimum percentage, while offering small increases in the federal matching rate to states that reduce interdistrict disparities by more than the minimum percentage.[72] This requirement of intrastate equalization would drive federal aid to the neediest districts and schools within each state, thereby subsuming the objectives of Title I. To enhance continuity with Title I, the program could specify that within-state allocations in accordance with the current district- and school-level allocation formulas of Title I would presumptively satisfy the intrastate equalization requirement.

In sketching the basic contours of a national foundation program, I recognize that, in the hands of Congress, all of the parameters--pupil weights, cost adjustments, minimum state effort, federal matching rate, and the foundation level itself--would be informed by a complex mix of research, expert judgment, and politics. The practical balance of benefits and burdens is as important as any distributive principle in determining the shape of a viable program. Nevertheless, as long as public demand for high standards can be sustained, and as we learn more from cost studies about current shortcomings in financing a truly adequate education, the case for a robust federal role in narrowing interstate disparities and ensuring a national foundation level of resources will remain strong.

To gauge the potential impact of this reform, I compared the interstate equalizing effect of federal education aid in 2002-03 with the effect of a program with the following parameters:

i. Foundation guarantee. The program assures every state at least $6500 in cost-adjusted revenue per weighted pupil, an amount that Congress has hypothetically determined, based on the best available evidence, to be a reasonable estimate of the cost of adequate educational opportunity for equal national citizenship.

ii. Minimum state effort. As a condition of federal aid, each state with nonfederal per-pupil revenue below $6500 must devote (a) at least 3.25% of its total taxable resources to education or (b) the level of effort necessary to produce the $6500 foundation level, whichever is less. In other words, a state is ineligible for federal aid if it has not made sufficient effort to bring its per-pupil revenue up to the foundation level.

iii. Federal matching rate. Each state's nonfederal revenue is matched by federal aid at a rate inversely proportional to the ratio of the state's fiscal capacity to the national average.

iv. Minimum matching rate. The minimum federal matching rate is set at four percent, a figure hypothetically judged by Congress to be high enough to garner support for the program from relatively wealthy states.

Table 8 simulates the results of this program. Column A shows cost-adjusted revenue per weighted pupil from all sources for each state in 2002-03, and Column B shows cost-adjusted revenue per weighted pupil from nonfederal sources.[73] Column C shows per-pupil revenue after applying the minimum effort requirement to states in Column B below the $6500 foundation.[74] Column D lists the federal matching rate for each state according to a formula that increases the rate as state fiscal capacity decreases, with a minimum rate of four percent.[75] Column E applies the matching rates to the figures in Column C to produce the total cost-adjusted revenue per weighted pupil for each state under the program.[76] The enrollment-weighted coefficient of interstate variation is shown at the bottom of the columns.

As the matching rates in Column D indicate, the simulated national foundation plan disproportionately benefits states with relatively low fiscal capacity that have exerted at least the minimum effort, such as Alabama, California, Idaho, Montana, New Mexico, and Oklahoma. The plan is less generous toward states with relatively high fiscal capacity, including not only states with historically high education spending, such as Connecticut, Massachusetts, and New York, but also states whose low education revenue is largely due to low effort, such as Florida, Nevada, North Carolina, and South Dakota. The plan thus ensures a base level of per-pupil funding by directing substantial aid to poorer states where additional money is likely to yield the greatest educational dividends, while encouraging wealthier states to do their fair share.

The parameters of the federal matching rate, foundation level, and minimum state effort can be adjusted to produce greater or lesser degrees of interstate equalization. The main point is that the program in its essentials is structured to deliver far more equality of opportunity across states than current federal policy. The program simulated in Column E would have narrowed interstate inequality in per-pupil revenue by nearly one-third (32%) at a cost of $43.5 billion in 2002-03.[77] By comparison, actual federal education revenue in 2002-03 totaled $36.8 billion and reduced the coefficient of interstate variation by only 12%.[78]

If Congress were to adopt this national foundation plan as a major reform and expansion of Title I, it would require approximately $30 billion in new money above the $13 billion currently spent under Title I.[79] Large as this increase may seem, it is consistent with other estimates of the cost of a national foundation plan,[80] and the federal share of the national education budget would still be less than 15%.[81] Moreover, a significant component of the $43.5 billion estimate in Table 8 is attributable to the four percent minimum federal matching rate. As Columns F and G show, the plan without any minimum would have produced an even greater degree of interstate equalization (a 37% reduction in the coefficient of variation) at a lesser cost ($37.2 billion) in 2002-03, although only thirty states--perhaps too few for an effective political majority--would have received significant federal aid.[82] Ultimately, any fair assessment of the desirability of new education spending must also take into account the social and economic costs of educational inadequacy.[83]

Conclusion

To be sure, the shortcomings of American public education are too complex and multifaceted to be remedied by simply "throwing money at the problem." The national foundation plan I propose must grow out of and bear a reasonable empirical relationship to learning standards that lend coherence and strategic direction to education policy in the area of school finance and beyond. Such reforms also must be nested within ongoing efforts to improve the accountability and efficiency of public schools. Moreover, districts and schools need concrete solutions to intensely practical challenges, such as how to provide teachers with sufficient time and professional development to align their knowledge and practice with higher standards, and how to implement and refine best practices for improving the performance of the most disadvantaged students. Given this context, the ideas presented here are not intended to be panaceas. To be effective, they must leverage and integrate other reform agendas in the policy environment.

At the same time, it is difficult to believe that our gaping interstate disparities in educational standards and resources have little or no bearing on unequal opportunity and outcomes. The problem is one that only the federal government can meaningfully address. The political alignment necessary for a solution is a topic beyond the scope of this paper. But the approach must bring together Southern moderates who see the benefits of federal assistance outweighing the threat to states' rights with Northern liberals who support a fairer distribution of the nation's wealth. Today the coalition might also include legislators from the West and Southwest, where high poverty and immigration have produced formidable educational challenges. The viability of any reform will of course depend on the balance of winners and losers. But without a new and concerted effort, it will continue to be more rhetoric than reality to speak of a national commitment to equal educational opportunity.