Building Back Better: Investing in Improving Schools, Creating Jobs, and Strengthening Families and our Economy

From the Coronavirus Response and Relief Supplemental Appropriations Act, to the American Recovery Plan Act, to the American Jobs Plan, to the information reported as of the time I am preparing my testimony on the American Families Plan, to President Biden’s 2022 budget proposal, it is clear that the president wants to direct considerable new dollars to education. My rough estimate is the increase in federal education spending would be in the neighborhood of $120 billion per year over the next several years, with different laws and plans having different timelines. Considering that in 2019 total on‐​budget federal support for education was about $246 billion, that is a very large increase.¹ We need to ask two key questions about such spending: (1) Is it constitutional, and (2) is it likely to be an effective use of taxpayer resources? I will briefly address (1) and spend more time on (2) based on what we can tell about previous spending increases. I will also address “free” college and improving physical conditions of schools — two aims of present, and likely upcoming, legislation.

The federal government has only specific, enumerated powers, primarily found in Article 1, Section 8 of the Constitution, and authority to spend money on education outside of federal lands or territories is not among them. Importantly, the “general welfare” clause does not authorize such expenditures. As James Madison explained in Federalist no. 41:

Similarly, Alexander Hamilton, writing about the taxation and “necessary and proper” clauses in Federalist no. 33, noted that the federal government is only given specific powers:

Finally, records of the 1787 constitutional convention include almost no mention of education save for some discussion of authorizing creation of a national university. Of course, the authority to create a national university has never been among the enumerated powers, but not due to an assumption that it would fall under the “general welfare” clause. No, because a different enumerated power could cover it. As James Madison recorded Gouverneur Morris of Pennsylvania explaining, enumerating “it is not necessary. The exclusive power at the Seat of Government, will reach the object.”²

“Is it constitutional” should always be the first question asked of federal legislation. But if satisfied that it is, likely effects of legislation should be next. So, will very large increases in spending produce commensurate improvements in education? Contrary to what may be a common impression, funding for American education has risen appreciably over time, and we can look at corresponding achievement results to attempt to gauge whether the country has seen commensurate improvements in learning outcomes.

Elementary and Secondary

In elementary and secondary education, according to inflation‐​adjusted federal data, total spending per public‐​school student in Fall enrollment has increased from $646 in the 1919–1920 school year — the first year listed in the federal Digest of Education Statistics — to $4,893 in 1965–66 — basically the beginning of major federal funding — to $14,891 in 2017–18, which is the last year with available data.³ During just the period of federal involvement, real spending has tripled.

What has happened to student achievement in that time? First a proviso: What people want out of education, and how to measure achievement, are much less clear than one may commonly assume. Some people think education is primarily about shaping character. Others, good citizens. Yet others, about providing students with the skills and knowledge to earn enough money to live comfortably as adults. Many of these outcomes do not lend themselves to clear‐​cut measures. Meanwhile, those goals that seem like they could be clearly measured — literacy, numeracy — are not easy to test reliably, with outcomes impacted by test wording, exam length, testing room conditions, consequences attached to test performance, and more. As a result, there is no single metric that can tell us how well our public school system is working.

That said, the federal government established the National Assessment of Educational Progress (NAEP) to conduct assessments of different types and on numerous subjects to gauge how the country’s K‑12 system is working. What those scores suggest — in particular the mathematics and English/​language arts scores that address the core of education — is that spending increases do not translate into commensurate, lasting improvements, with the proviso that “commensurate” is a subjective term; your mileage may vary.

Most basically, we have seen very little movement on the average scores on the Long‐​Term Trend (LTT) NAEP exam for 17‐​year‐​olds, basically the “final products” of the nation’s elementary and secondary education system.⁴ The LTT endeavors to keep the test consistent from its first to its most recent year in order to have a seamless measure of achievement.

The average score in math in 1978 — the first year the exam was given — was 300 out of 500.⁵ As of 2012, the last year the exam was given, it was only 6 points higher, at 306. In reading the results are even less encouraging, with the average score in 1971 at 285, and in 2012 only 2 points higher at 287. Between 1959 — 12 years before the first LTT reading test, which captures the 17-year-old’s full education — and 2012, real per‐​pupil funding rose from $3,852 to $13,554, a 252 percent increase.

To put the scores in context, the LTT also identifies performance levels with cutoff scores. The second highest is a score of 300 and above, and seeing the change in the share of students surpassing it may give a slightly different perspective on changing achievement. Here the news is only slightly better than averages, and again it seems very hard to say we have gotten much bang for our buck. In 1978, 52 percent of 17‐​year‐​olds met or exceeded the 300 mark in math. By 2012 that had risen to only 60 percent. In 1971, 39 percent hit 300 or above in reading, and the exact same share hit it in 2012 — no improvement for more than tripled funding.

More recent than the LTT scores are “Main” NAEP outcomes. These exams are not intended to be consistent going back to the 1970s but are given more often and provide more recent trend data. They also designate scores as demonstrating “basic” level mastery, “proficient” mastery, and “advanced” performance. The test is also aligned to grade, not age, though there is significant overlap in those two things.

Unfortunately, 12th grade math scores only exist between 2005 and 2019, but the story is similar to the LTT. The average math score in 2005 was 150 — in this case out of 300 — and it was 150 in 2019. No change, while spending rose from $10,472 in 1993 — twelve years before the exam — to $14,891 in 2017–18, again the most recent year with data. In reading, scores go back to 1992 and the average dropped from 292 out of 500 to 285. “Proficiency” levels were no more encouraging, dropping from 61 percent proficient in math in 2005 to 60 percent in 2019, and from 40 percent to 37 percent in reading.

Rising spending and dropping performance is clearly not a positive sign for the effect of spending.

Of course, federal elementary and secondary spending is aimed at low‐​income students. How did those scores change?

Looking at the LTT, we cannot pinpoint students’ family income, but can approximate it by breaking scores down by parental education levels and focusing on students with parents who did not finish high school. We will look only at the share of students passing the second‐​highest scores threshold, which is a bit more concrete than examining average scores.

In math, in 1978, 26 percent of students with parents who did not complete high school equaled or passed the 300 mark. In 2012 that was up to 37 percent. That 11 percentage point increase is not trivial, but it is hard to say that it is commensurate with overall spending more than tripling, and federal funding going from $481 in 1969–70 — as close as the data gets to 12 years before 1978 — to $1,391 in 2012, which is also nearly a tripling.⁶ In reading, results by parental education only go back to 2004, and the share of our focus students meeting or exceeding 300 rose from 17 to 19 percent. Overall per‐​pupil spending in 1992 was $10,472, while the closest year with federal spending data was 1989–90, when it was $635. So overall spending increased from $10,472 to $13,554, or 30 percent, and federal spending rose from $635 to $1,391, or 119 percent — a lot for a 2 percentage point uptick.

On the Main NAEP we can again proxy low‐​income scores by looking at parental education levels. In math, 7 percent of 12th graders whose parents did not finish high school were “proficient” in 2005. That rose to 9 percent in 2019 but the increase was not statistically significant. Meanwhile, overall spending rose from $10,472 in 1993 to $14,891 in 2017–18, and federal funding from $720 to $1,175. In reading there was no statistical difference between 1992 and 2019, with the share proficient dropping from 21 to 20 percent. Overall spending was $7,586 in 1980 — roughly half the amount in the most recent year — and federal spending was $756, about two‐​thirds of spending in the last year.

Based on these outcomes it appears we have bought very little meaningful improvement with taxpayer dollars. While spending has grown appreciably, results have essentially stagnated and in some cases even declined. Importantly, we have seen better outcomes in lower ages and grades, and it may be there is something peculiar about 12th grade test‐​taking. Maybe, for instance, the students just do not care about the test. But unless the level of not caring increased over time we should still see improving outcomes were increased spending an ultimately positive force.

One other thing is important to note: through much of this period we have seen marked increases in household incomes for poorer Americans once transfer payments and taxes paid are considered. Analyzing an October 2020 Congressional Budget Office report, the Cato Institute’s Scott Lincicome found that real income for the lowest quartile of earners rose from $19,300 in 1979 to $35,900 by 2017.⁷ This substantial improvement in resources should have had a big positive influence on scores, and may explain more about gains in all age categories than spending.

Higher Education

While we do not have national assessments in postsecondary education as we do in elementary and secondary, it may be more reasonable to lay outcomes at the feet of the federal government in higher ed than K‑12. While state and local governments certainly play major roles in higher education, establishing and funding public institutions from community colleges to large research universities, Washington provides a large share of funding overall via, especially, federal student aid programs including loans and grants, as well as some institutional aid and major research funding. In 2019 the federal government provided over $91 billion in student loans, $41 billion in research funding, and more than $107 billion in other funds including Pell Grants and aid to institutions.⁸ In 1965, in contrast, there were essentially no federal loans, and between research and other funding the federal government supplied only about $23 billion in 2019 dollars.

It is reasonable to conclude that hugely increased federal funding, especially student aid that became widespread in the 1960s and 1970s, helped to increase college enrollment and degree attainment, though it also fueled major price inflation.⁹ In 1960 only 7.7 percent of Americans 25‐​years and older had a bachelor’s degree or higher. In 2019 that number was 36 percent.¹⁰ The important question is whether this was a net gain for society.

Of course, rampant tuition inflation is a major problem. If much of your aid is burned off in higher prices it has done little good, and one recent estimate found that every dollar in increased loan aid translates into about 60 cents of increased prices.¹¹ But does the increase in degrees at least represent a significant increase in human capital — many more people able to do many more things that they lacked the knowledge and skills to do before?

We do not have a great deal of evidence on this, but we do have two sets of adult literacy assessments over time that enable us to see the average “literacy” — including the ability to read and comprehend, as well as “do math” — for people with various levels of formal schooling, including bachelor’s and advanced degrees. As spending and degree attainment rose, we should have seen scores on these exams holding steady or rising to indicate clearly expanding human capital. If they did not, it suggests that we have been producing more pieces of paper called “degrees” but which have decreasing substance behind them.

The first exam is the National Assessment of Adult Literacy (NAAL), which assessed adults’ ability to comprehend prose such as newspaper articles or brochures, documents such as tax forms, and quantitative literacy.¹² It was administered in 1992 and 2003, during which time the share of Americans ages 25 and above with bachelor’s degrees rose from 21.4 percent to 27.2 percent.¹³ We do not have readily available funding data for those exact years, but the closest available — starting in 1989–90 and ending 1999–00 — show real higher education spending per student rising from $20,572 to $28,549.¹⁴

Unfortunately, high degree attainment and spending were accompanied by decreasing literacy for degree holders. The share of adults who ended their formal education with a bachelor’s degree who were proficient prose readers in 1992 was 40 percent. By 2003 that had dropped to 31 percent. For document literacy the shares dropped from 37 percent to 25 percent. At least in quantitative results were unchanged, with 31 percent proficient in both years. The direction for adults with advanced degrees was also bad, with those who were prose proficient dropping from 51 percent to 41 percent, document from 45 percent to 31 percent, and quantitative from 39 to 36 percent, though the latter was not statistically significant. The NAAL results pointed toward credential inflation — more sheepskins rather than more human capital — and money poorly spent.

The second exam, which has essentially replaced the NAAL, is the Program for the International Assessment of Adult Competencies (PIAAC), which was administered in 2012/14 and 2017. It is not directly comparable to NAAL, including that it does not designate “proficiency,” but it does give us performance levels akin to the NAEP LTT.

In terms of enrollment and spending, between 2012 and 2017 the share of Americans 25 and older with a bachelor’s degree rose from 30.9 percent to 34.2 percent, while higher education revenue per‐​student increased from $28,572 in 2009-10 to $34,606 in 2017–18. The literacy trend moved in the opposite direction. For U.S. Households with members ages 16 to 65 years old, in the 2012/14 administration 68 percent of people with more than a high school education scored in the third literacy level or above. In 2017 only 64 percent did.¹⁵ In numeracy the drop was from 57 to 53 percent. More movement in the wrong direction, though the PIAAC drops fell short of statistical significance.

That higher credentials have become increasingly empty as they have become increasingly numerous is corroborated by more than just assessments of adult literacy, including such measures as time students spend studying. In 1961 full‐​time students spent 25 hours per‐​week studying, in 1980 it was 20 hours, and by 2003 it had fallen to 13 hours.¹⁶ Add to this drops in median annual earnings of full‐​time, year‐​round workers ages 25 to 34 with bachelor’s degrees and above between 2000 and 2018,¹⁷ as well as long‐​term underemployment of about a third of four‐​year degree holders,¹⁸ and the indicators are powerful that we have massively over produced diplomas with our additional spending and enrollment.

In light of the data we have clearly indicating overconsumption and hollowing out of higher education, there is substantial reason to be concerned about “free” college proposals. Such proposals vary in their specifics — they can include government directly funding colleges so they charge no tuition, no tuition and fees, or even government funding schools directly and supplying students money for shelter and food — but all have the goal of reducing the amount students pay for their education.

Such proposals are well intentioned, especially as one considers the astonishing sticker prices at some colleges and universities. But a root problem remains no matter whether government supplies aid to student or funds colleges directly: When the consumer does not pay the price with their own money, or money they receive voluntarily from others, they will tend to overconsume education and direct more of their resources toward non‐​educational pursuits — partying, or frills such as on‐​campus waterparks¹⁹ — instead of efficiently focusing on the education they need to increase their earnings or obtain other core educational ends.

Subsidy‐​fueled over‐​credentialing also enables employers to increasingly demand degrees that may signify little about a person’s ability to do a job but that are often easy, basic screens for employers to weed some people — those who do not even have increasingly easy to get degrees — out. Indeed, research suggests that just such adding of diploma requirements to unchanged jobs has occurred.²⁰ That literacy exams have shown decreasing human capital for degree holders also points to the nation’s primary higher education problem not being that college is too expensive — though sticker prices are too high — but massively overconsumed, while putting everyone in a vicious cycle: the value of the average diploma is declining as credentials are watered down, but that means if you do not get a credential you will be increasingly behind.

Other things equal, making college free would do little to ameliorate the massive overconsumption problem. Indeed, it would likely make it worse, eliminating any of the discipline‐​inducing requirement that consumers pay for school at least using some of their own money. It may, though, decrease frills, as schools stopped competing for paying customers. But the same loss of impetus to provide frills could have bigger and more negative impacts.

A need to attract students, while hugely distorted by subsidies to those students, is overall a good thing, driving schools to provide better and better experiences for students, including pleasant campuses, relatively easy access to professors, and more. Making college responsive, essentially, only to government, would change incentives from satisfying students to lobbying and navigating bureaucracies. It could also lead to rationing, as institutions would find themselves without the resources to expand and accommodate greatly increased demand. A recent study of OECD countries found that there are, indeed, negative correlations between subsidy levels and attainment — the greater the degree of government subsidy, the lower the rate of college attainment — and negative relationships between subsidies and higher education resources.²¹ Basically, more subsidization of schools is associated with fewer people completing and less well‐​resourced institutions.

The good news is this might alleviate the diploma glut. But it would do so at the expense of a higher education system that is currently very responsive to students and, compared to postsecondary education in the rest of the world, very dynamic.

Community colleges, which are the least expensive sector of American higher education, are relatively easy to make free to students, with average tuition and fees of $3,377 in 2019–2020.²² To put this in perspective, the average Pell Grant in 2018–19 for an undergraduate student was $4,418.²³ Of course, students also need food and shelter, but would have those expenses whether they were students or not.

What we see in the low‐​cost community college sector, however, is what may well be a lot of poorly focused schooling, though there may also be an effect of low resources and quality as the sector aims for a low‐​cost model. Whatever the reason, completion rates for community colleges are very low. According to data from the National Student Clearinghouse, which has data on about 97 percent of total college enrollment, of the cohort of students who started college in 2014 and did so at a two‐​year public college, only 40.2 percent had completed a program of study within six years.²⁴ To put that in perspective, 76.7 percent of students who had started at a four-year not‐​for‐​profit private school had completed their program.

Of course, making public colleges free would hurt, and likely kill, many private colleges — often religious in nature, which public institutions cannot be — which would find competing against “free” impossible.²⁵ Except, that is, for elite institutions such as Harvard, Stanford, Yale, and other rare institutions with big names and large endowments, which would become even more preserves for the rich as others swarmed free schools. Of course, were free public colleges to continue to be allowed to be selective in enrollment we would still see elitism beyond the wealthy being able to pay for high‐​profile private institutions, including wealthier students likely in K‑12 systems in which there is more knowledge about how to work in an increasingly bureaucratic system to gain entry to preferred schools.

Free college would make higher education less expensive for students, and perhaps for society, if it were to replace subsidies to students. But the effects would almost certainly be overall losses, as we either produced more credentials in an already glutted market, in the process requiring even more credentialism just to stay in one place, or forced rationing which would likely favor the well‐​connected and maybe still not reduce the glut to a reasonable level. And it would almost certainly sacrifice quality in a system which, for all its serious flaws, dominates lists of top institutions in the world.²⁶

School Conditions An emphasis of the Biden administration is fixing the nation’s infrastructure such as roads and bridges, to include school buildings. According to a 2020 GAO report, about 41 percent of districts report that at least half of their schools need updates or replacements of the HVAC systems, about 28 percent of their interior light fixtures, and more.²⁷ That may seem like a desperate situation, but the available evidence suggests that there is no crisis of crumbling schools, and that major federal aid dollars would not largely be used to repair important, but hidden and dull, items such as HVAC units. It would be used for flashy things schools do not list as in need of replacement or repair, such as purchasing and installing new technology.

First, the overall condition of schools may not be especially bad. As recently as the 2012–13 school year, a federal report found school districts reporting that only 3% of permanent buildings were in “poor” condition, meaning they fell short of “minimum requirements for normal school performance.” That rose to 9% for portable buildings. Even in poor districts — those with at least 75% low‐​income students — only 4% of permanent buildings were reported to be in poor condition, and roughly 8% of portables.²⁸

Other data suggest that districts tend to use facilities money for building new schools, which can again be “flashy” projects that draw a lot of positive public attention. A survey of readers of the journal School Planning and Management found that 59 percent of districts in 2019 completed some sort of construction project, including nearly one‐​quarter competing construction of new or replacement buildings.²⁹ 58 percent planned to start new construction projects in 2020. It also seems that, when asked to pay for their own infrastructure, communities are hesitant, with the survey finding that a commonly reported impediment to construction is “community support to pass a bond referendum.” People tend to be strict when their money is involved.

The GAO also reported findings suggesting that districts are more willing to put money into high‐​profile items like technology than nuts‐​and‐​bolts such as boilers. “Student access to technology” were districts’ second highest priority, after “safety and security (e.g., cameras, alarms, access control),” despite the fact that employers are looking for “soft skills” like good communication rather than technological prowess, which young people tend to have as “digital natives.”³⁰ As GAO investigators reported of a Rhode Island district, “Officials said participants in public forums told them they preferred educational enhancements over facility repairs.”³¹ Officials in other districts talked about having to bundle more “fun” stuff, like equipment for robotics labs, with HVAC repairs to get the latter passed, which is an inefficient use of funds. Finally, new schools do not tend to be simple replacements, but they keep getting bigger, with data showing that between 1995 and 2014 space increased by 30 square feet for each high school student, 45 square feet for each middle school child, and 80 square feet per elementary school child.³²

It is not clear that public schools in general are in seriously poor condition, nor that were they to receive large sums of federal money it would be used to address primary, but unglamorous, problems, like updating HVAC systems. It is also worth noting that in contrast to understandable predictions at the beginning of the COVID-19 pandemic, school districts have likely not taken major financial hits. Indeed, state and local tax revenues were higher in calendar year 2020 than 2019, and some states and districts are struggling to determine how to use the federal windfall through the three COVID-19 relief bills: CARES, CRRSA, and ARPA.³³

The desire to put as much money as possible into schools at all levels is understandable. Education is generally a good thing, and other things equal, when we spend more on something we get more of it, higher quality, or both. But existing evidence suggests that increased spending in the past did not translate into commensurate achievement gains. In elementary and secondary education more spending tended to coincide with small if any achievement gains for those at the end of K‑12 schooling, including for the low‐​income students federal money is supposed to target. At the higher education level, where the federal impact is much greater and, hence, more clear, greater spending would likely create more credentials but less learning per credential, while fueling a vicious cycle of credentialism that forces more and more people to spend precious time in school without much learning. And as we have seen, the more people use other people’s money for things, the less efficient expenditures tend to become. For these reasons, and because the Constitution does not authorize any education spending outside of federal lands, Congress should steer clear of major increases in education spending.