Matt Yglesias argues that we rarely think through the implications of Baumol’s cost disease:

For example, suppose you’re thinking about spending on teacher compensation. The following options are available:
— Class size can steadily increase over time.
— Teacher pay relative to average pay can steadily decline over time.
— Taxes can steadily rise over time.
As of January 2010 only four percent of people thought public school teachers are paid too much. And yet the public is hardly clamoring for larger class sizes or endless tax hikes.

Matt makes a number of solid points, but I’d add one wrinkle. As (ahem) Rick Hess has observed, student-teacher ratios have been decreasing (some would say improving) markedly in recent decades, yet we haven’t seen commensurate gains in educational productivity. Had we remained at the student-teacher ratios that prevailed four decades ago, we’d have 2.2 million public school teachers rather than 3.2 million. The teaching force has grown 50 percent faster than student enrollment. That is an eye-popping number.

So we could have:

– kept class sizes fixed — not increased them! — and dramatically increased salaries for all teachers while spending the same amount, improving the teacher talent pool and educational outcomes while keeping spending the same

– allowed the best teachers to teach more students in exchange for more pay, improving the talent pool and outcomes while reducing spending

There are many other scenarios we can easily conjure up, e.g., what if work rules didn’t prevent new collaborative forms of teaching that could deploy existing staff more effectively by facilitating specialization, etc.

As for the polling result, it reflects the fact that teachers are one of the most well-regarded professionals in the U.S., thus enhancing their political effectiveness. I do wonder about the kind of answers we’d get to other questions: should we pay the best teachers much more than others, should school administrators have the power to pay teachers the way companies pay their employees, etc. I don’t know what the answers would be, but I think they’d tell us more than answers to the question Matt cites.