It’s surplus education, stupid

In 1970 America spent about $228 billion in today's dollars on public schools. In 2007 that figure was $583 billion. True, some of the increase can be traced back to growing enrollments, better programs, and improved services for special-education and other students, but much of the increase is just a lot of spending without a lot to show for it. — Andy Rotherham in Time.

I'll have more to say about this argument during my winter break, but an essential dilemma of publicly-funded schooling is that much of it is surplus no matter how you want to slice the issue. As critics of reports from A Nation at Risk to Rising above the Gathering Storm have pointed out over the years, it is hard to point to generalized human-capital deficits that are commonly asserted: while there are spot shortages in areas almost any year (including in this economic climate), that's different from arguing we have a generalized [missile] [mine-shaft] engineering gap that's an immediate crisis. 

That criticism becomes obviously and statically wrong only if you realize that in the U.S., educational attainment has historically run well ahead of any rationalized present labor-market demand. From a contemporary labor-market perspective (at the time), it was completely useless to have coeducational primary schooling in the early 19th century. Or to have a dramatic expansion in high school attainment among African-American Southerners in the 1940s and 1950s. And the common rationale for tracking in the early 20th century was precisely the type of "well, they're not going to get the best jobs so let's train them for the type of jobs we expect them to have" cramped perspective that some of my academic friends have unfortunately echoed with a more egalitarian and late-industrial slant. 

The common non-economic arguments for providing more sensible and equitable educational access:

• As a preparation for democracy, a minimum of education must be provided to all citizens. 
• As a public good, education must be available on an equitable basis. 
• As a public service outcome with private benefits, access to educational credentials must be available to all. 
• To prepare the next generation of teachers, we need to provide advanced education to a significant minority. (This was really W.E.B. Du Bois's point in the "Talented Tenth" argument he made. It was not about elites; it was about teacher education.)

In addition, you can make a reasonable argument that putatively-surplus education in and of itself will pay off in many (and often unexpected) ways. But that's a general argument, and it doesn't tell you how much surplus education is good or how to balance that against other ways of spending public resources. 

As a share of GDP, my impression from a handful of sources is that elementary and secondary education spending has remained reasonably constant over a few decades, while higher-ed spending has increased (and I suspect a disproportionate part of the last decade's increase has been in the for-profit sector of higher ed). So the common complaint that we're not getting more bang for the buck (which Rotherham echoes) is more applicable to higher education than elementary and secondary education. Yet that's where the surplus-education argument is most applicable. 

My instinct is that the increase in per-pupil spending at all levels and the %-GDP increase in higher ed is driven by a few factors:

• An increased effort (often mandated) to serve students with disabilities and other populations who require more specialized services.
• Increased overhead in a death-of-a-thousand-cuts style–indirect increases in maintenance costs from greater capital expectations (state code requirements, parent expectations for what a "modern school" or "modern college" should have, etc.), the support required for generating IEPs and all sorts of other compliance-related documents, supervisory support, and various indirect services such as speech and language therapy, high school advising, security on campuses. 
• Increased unionization since 1970 in both K-12 and higher education (public sectors). The shiny-object attention right now is on stuff like pensions and masters pay bumps for licensed teachers, but I suspect there's been a measurable but undiscussed cost in paying even halfway-decent wages to teachers' aides, cafeteria workers, bus drivers, maintenance staff, and so forth. And I suspect that few who would want to focus on pensions want teachers' aides to be homeless. 
• Rent-seeking behavior by textbook, journal, and test publishers.
• Class-size reductions in some areas (which are very difficult to quantify accurately)

What's maddening to me is the parenthesis in the last bullet above: anything like cost accounting is virtually impossible in a large interdependent bureaucracy (many K-12 systems) and given the extent of cross-subsidies in higher education. Can you spin this? Sure: some part of this is what Corey Bunje Bower has called luxury consumption, and that makes some sense from a long-term perspective. But few are going to agree with my long-term focus in the next few years, and this presents a fundamental political dilemma: how to address the cost issue when the next few years' state budgets are going to be very, very cramped. Rotherham's essay is not the first and is not going to be the last to make this bang-for-the-buck argument.

(Yes, I'm a bit disappointed that Rotherham gave the raw-numbers cost figures instead of focusing on per-pupil expenditures or percentage-of-state-budget or other figures. But I assure you that fully-adjusted figures would only moderate and not eliminate the issue.) Update: Bruce Baker has more. Baker's chosen measure is education spending as proportion of personal income, and for that measure, using Tax Policy Center data, he points to considerable interstate variability in trends. Given the different structure of tax bases by state, I'm not sure that education spending as a proportion of personal income is the best measure. In addition, state figures for personal income can be skewed away from other economic data by residence of wealthy individuals and by non-wage income. But the variability point is going to hold with other measures.