Commentary | Wages, Incomes, and Wealth

Jay Greene’s persistent misuse of data for teacher pay comparisons

Opinion pieces and speeches by EPI staff and associates.


Jay Greene’s persistent misuse of data for teacher pay comparisons

By  Lawrence Mishel, EPI President

Jay Greene’s new research on teacher pay claims to show that teachers are well-paid when one looks at their weekly or hourly wage compared to other professionals. This is pretend social science, at best.

Greene’s conclusions are solely based on the only data (National Compensation Survey) that one could point to that show teacher pay above average; he willfully ignores all other data and even all other studies of teacher pay. The use of the NCS data to compare teachers to other professionals using hourly or weekly wage measures is very controversial. This is because these data are fundamentally flawed and inappropriate to use in this way. My co-authors and I showed why these data are flawed in a study released in 2004, How Does Teacher Pay Compare? (pp. 30-39). We even present quotes from the chief Bureau of Labor Statistics statistician overseeing these data saying the use of these data for this purpose is inappropriate (see below). Michael Podgursky, one of the few researchers besides Greene who believes teachers are well paid, has stopped using the NCS hourly wage measure, though he continues to use the weekly measure (both are flawed, though the hourly measure is doubly flawed, as I will explain). The controversy over these data was aired in a debate that was given a fair amount of attention.

Let me explain why we said in How Does Teacher Pay Compare that “hourly or weekly wage data from the NCS (as given) should not be used to make comparisons between teachers and other occupations.”

The bottom line is that these NCS data are based on employer surveys and the NCS measures scheduled   hours. Because teachers do not work the familiar full-year and roughly 9-to-5 schedules that most professionals have, the comparisons is one of apples to oranges. Here is what the BLS statistician wrote after giving an example:

The lesson of this example is the futility of comparing salary estimates for periods of less than a year between two occupations with very dissimilar work time requirements over the year and for whom the annual leave entitlements are unknown.  In our example the weekly salary difference changed from 140 to 124 percent.  Because the published NCS wage estimates do not reflect leave entitlements and the work years of teachers are so dissimilar from most other professional occupations, I would only use the annual salary estimates from NCS to compare teacher pay with the pay of other professionals.

Yet, it is precisely these data, and only these data, that Greene relies on. The bottom line is that the Manhattan reports weekly and hourly wages which rely on information that vastly understates the weekly hours of teachers and the weeks teachers work each year, and thereby significantly overstate the hourly wage or weekly wage for a given annual wage. Let me explain in more detail.

1. Because the weeks that teachers work per year are understated, their weekly wage is overstated.
The NCS relies on the number of official school days to calculate weeks worked (note that weekly wages is annual wages divided by weeks worked). The NCS reports that teachers work 38 weeks each year. However, the NCS reports that other professionals work 51 or 52 weeks (see Table 11of How Does Teacher Pay Compare?). This is strange but understandable if you know how the survey measures time worked. Teachers are measured by days worked (say 190 official school days divided by five, resulting in 38 weeks)m while others are measured as days paid (work days plus paid time off: breaks, vacations and holidays). How else can one understand other professional occupations working 51-52 weeks a year? Do they not take Federal holidays (10 of them), enjoy other holidays (day after Thanksgiving), or have paid vacations or personal leave? Teachers have more than a 38 week work year, unless you believe the summer off lasts 14 weeks (52 less 38), which is saying teachers have more than three months off in the summer. The issue here is that these data don’t count teacher holidays and vacations during the school year (spring and winter breaks, etc.) when measuring weeks, and the data don’t take into account paid leave when calculating weeks worked for professionals. To be comparable, the occupations would have to have their weeks counted the same way, either by paid weeks or by scheduled /worked weeks.

2. Because teachers’ hours worked per week are understated, hourly wages are overstated.
So, teachers are considered to work 36-37 hours per week in the NCS, as measured by the official school day, less lunch, etc. This reflects their scheduled work hours.  Other occupations don’t have a regular rigid schedule and employers report that accountants, reporters, etc. work, not surprisingly, about 40 hours (again, see Table 11 in our study). There are several studies that show that teachers work as many, if not more, hours than other white-collar workers. The hours that teachers work at school, either before or after the bell, or their work at home are excluded  in the NCS measure. The hourly wage measure is doubly flawed because it relies on dividing a flawed measure of weekly wages (see point #1 above) by a flawed measure of weekly hours.

If one corrected for the understatement of teachers’ weekly hours and weeks worked per year, then you wouldn’t find teachers are “well paid.”

Further explication from the BLS

Given these inconsistencies, it is not surprising that the BLS chief of the Division of Compensation Data Estimation National Compensation Survey noted, in correspondence, that:

We are actively studying which occupations and groups of occupations should not have earnings estimated and published by hour.  Flight crews on airlines, outside sales representatives, operating personnel on over-the-road railroads are but a few of the occupations in addition to teachers.  The point of this research is to define comprehensive requirements for the estimation systems changes contemplated for 2007.  While we cannot change the tables before 2007, we may be able to amend the introduction sections to our local and national bulletins to include a cautionary note in the near future.

A bit of fun: If you believe the NCS Hourly Wage Data*, then you believe:

Economics professors (


) are the second highest paid professionals after pilots and make far more than business professors (



English professors (


) make more than dentists (


) or nuclear engineers (



Law professors ($51.71) make the same as judges (


) and more than doctors (



Physical education professors (


) make more than architects (


), accountants (


), and computer programmers (



Sociology professors (


) make more than mechanical engineers (


) and architects (



* Based on wage data for 2002.

Lawrence Mishel is president of the Economic Policy Institute in Washington, D.C.


See related work on Income and wages | Wages, Incomes, and Wealth

See more work by Lawrence Mishel