Does paying teachers more (relative to what other college educated workers earn) lead to better results. Today, a chart showing reading scores (PISA scores) in 25 countries and how they vary with how well teachers are paid.
Given that I spend a good part of my day training students in econometrics, I want to raise the flag that reads “These results are interesting; they’re not carefully vetted science.” More on that below for those interested. First, the results.
The vertical axis measures reading achievement for 15-year olds in a given country. The horizontal axis measures teacher salaries relative to other college-educated workers. The conclusion is higher salaries do lead to better performance. Looking at the slope of the red line suggests that if U.S. salaries were comparable to those in Finland, we’d close about a quarter of the test score gap with Finland. Similarly, if our salaries matched Korea’s, we’d close half the gap with our friends there.
Higher teacher salaries won’t solve all our educational problems…but they’d be a real good start.
Now for all the scientific caveats. What I did was take teacher salary data from the OECD’s Education at a Glance, 2012 and 2009 PISA reading scores from Wikipedia. I used all the countries for which both numbers are available, except Luxembourg (which is an outlier in many ways.)
The t-statistic on the regression line shown is 2.3, which is plenty significant. As a robustness check I included GDP per capita in the equation. Nothing much notable changed. Having said that, there are only 25 date points in the sample and there are no doubt other variables that I haven’t controlled for. So while I believe these results are interesting–you wouldn’t be seeing them otherwise!–they’re certainly not definitive.