A lecture within “Modern Economic Issues”, a course by Robert Whaples, concerns the productivity of United States public schools. The overall statistic is fairly familiar. Over the thirty years between 1970 and 1999, spending on public schools roughly doubled in inflation-adjusted dollars. The number of students and their ability, as measured by standardized tests, remain constant. Since productivity is simply output divided by cost, this means that productivity fell by 50 percent. (There are plenty of excuses made by teachers’ unions for this drop, e.g., that immigrants and non-white students are harder to educate, but economists have found that these factors should be canceled out by the fact that today’s parents are better educated than the parents of children in 1970.)
It occurred to me that actually productivity for public employees such as teachers is impossible to measure. Unless we get a letter from God saying how long each teacher will live after retirement, there is no way to know how long retired teachers will live and therefore there is no way to estimate the pension and health care costs. Here in Cambridge, Massachusetts, we know that out of $26,305 spent per student in 2011, $7004 went to pay teachers while $5051 funded “insurance, retirement”. The total budget, however, is understated for the obvious reason that it does not include the capital cost of the schools themselves (“capital expenditures and debt service payments are excluded from this calculation”) and the non-obvious reason that, to the extent that pension costs are included, they are based on the hope that investments in stocks will yield an 8 percent return as well as a prediction on how long today’s 35-year-olds will live.
Then I realized that U.S. productivity statistics are probably inflated artificially over those of countries such as Holland, Singapore, Australia, etc. that have funded pension systems and government-budgeted health care costs. Private sector U.S. workers today participate in Social Security and Medicare, whose future costs are unknown but almost surely vastly higher than the amount being collected in tax from current workers. So the U.S., despite its bountiful resources and fully developed infrastructure, may be a lot less productive than current statistics show.
Whoa, whoa, whoa. Productivity isn’t output divided by cost! Is that what Whaples said? It’s the value of output per hour worked (either in goods or services)
Productivity is vitally important because it determines overall living standards (across the whole economy, you can’t consume more than you produce).
You improve productivity by consuming less than you produce, and investing in capital (you can dig ditches much faster with a backhoe than with shovels). This is why rich countries are rich–they have much more capital than poor countries.
Did Whaples talk about Baumol’s cost disease? For labor-intensive occupations (like teaching), it’s hard to raise productivity (what could you invest in that would reduce classroom hours by 25%, or even 10%?). But wages depend on average productivity across the whole national economy. So output divided by cost will go down.
Russil: I don’t think that Whaples was talking about “labor productivity”, which is conventionally defined as output per hour (just as you say). He was talking about productivity in general, which I think is generally defined as output divided by input. The easiest input statistic for economists to get hold of is presumably school budgets (dollars) rather than hours worked by school employees. Anyway, if you’re a taxpayer trying to decide if your money would be better spent on something else you would be more interested in the dollars spent. Measuring the dollars spent captures things like school systems spending on IT or central administration.
Whaples did not talk about Baumol.
Thanks, Philip. I believe the usual productivity stats used for international comparisons are labor productivity and unit labor costs (hourly compensation divided by hourly output). In these terms, if I understand correctly, what you’re basically saying is that unit labor costs (plus capital spending) are too high, and probably underestimated.
Paul Krugman has an interesting blog post on this subject. Is the problem that labor costs are too high? Or is it that real interest rates are too high?
Russil: I certainly did not suggest that “unit labor costs are too high”. I offered no opinion on whether wages were too low or too high (I’m not even sure what that would mean, except that for my own personal wages I would like them to be 10X higher). My point was that the dollar output per dollar of labor input stats in the U.S. might not be comparable to those in other countries and therefore not as useful as they would appear for assessing competitiveness. We have pushed off so many costs into the future that we might be fooling ourselves into thinking that the U.S. is a sensible place to do business. Then we will be bewildered as companies choose to invest and expand in countries other than the U.S.
Sorry, maybe I wasn’t clear. “Unit labor costs” (cost divided by output) is just the reciprocal of what you’re calling “productivity” (output divided by cost). If you think the US isn’t competitive because its productivity A is too low, that’s the same as saying that it isn’t competitive because its unit labor costs 1/A are too high. If everything else is equal, an employer will choose to hire in a country with lower unit labor costs.
The advantage of talking about unit labor costs is that they’re measured and published every year, for a large number of countries. See page 39 in this 2012 report, for example.
The compensation measured as part of unit labor costs doesn’t just include wages: it also includes benefits, labor-related taxes, and contributions to social benefit funds like unemployment insurance and pension funds. (It doesn’t cover everything, though: as you noted, if a defined-benefit pension fund is running a deficit, that results in an underestimate of the true cost.)
It’s worth noting that many companies are avoiding pension costs by using 401(k) plans instead. General Electric will be paying pensions to its current retirees for decades. Cisco Systems matches a portion of your 401(k) contributions, but other than a health care benefit for retirees, they don’t have to pay much for you after you’re gone.
Russil: You are basically saying that you don’t like the statistic that the economist lecturer used. He was very clear about his methodology. We spend twice as much, in real dollars, on education and we get the same number of students out, with no evidence that they are any better educated. But it was his lecture and he was presumably free to use this statistic. A lot of other economists apparently look at things the same way, i.e., how does output relate to costs. I was pointing out that when they do so the broad cost is understated because they aren’t accounting properly for pension, Social Security, and Medicare. And this may be why so many economists are scratching their heads wondering why the U.S. economy is growing slower than they expected.
Mark: The use of defined contribution plans means that the numbers for a private sector employee aren’t quite as distorted as those for a public employee, but it doesn’t change the fact that when costs for private sector employment are aggregated they understate the true cost of American workers. Eventually those workers will be over 65 and will need a huge amount of extra taxes to support. Potential investors in the U.S. know that those taxes will ultimate reduce their return on investment.