On school spending and teachers pay

Interesting.  Education, and schooling attainment, seems to be a big issue over here in NZ as well – with a political consensus that we care about this issue, but conflict about “how” school quality relates to this (or is even measured).  So lets have a chat.

Take schooling attainment as a measure of human capital, improved attainment suggests greater human capital attainment and income, and inequality in this measure as a measure of inequality in human capital  – when thinking about peoples opportunities, these types of inequalities are of pretty significant interest.  What role does school quality, or characteristics that are related to school quality, have on these schooling attainment outcomes.

In this paper the “school quality” variable is a function of two things: the teacher to pupil ratio and a measure of the average level of teacher salaries.  Fewer pupils per teacher may improve student outcomes as they get more teacher time, higher teacher wages may improve outcomes by attracting more qualified teachers or increasing the effort of the marginal teacher – hey it might not do these things, but the goal here is to consider what the data is saying first 😉

The three effects in the paper from considering this school quality variable to analyse educational outcomes?  Well copy and pasting from the paper we have:

  1. School quality is strongly related to schooling attainment for most parental education groups, with effects that are largest for children with the least educated parents and smallest for those with the most educated parents. The general pattern suggests that higher school quality contributes to a closing of between-family gaps in human capital.
  2. Estimated effects of the individual school quality measures are only slightly attenuated when we fit a model that includes both (teacher to pupil ratios (PT) and average salaries (W)) reflecting the limited correlation between PT and W across states.
  3. Estimates of teacher effects are little affected by the addition of several state-level controls—the average level of education of whites aged 25–55, the state-level white male unemployment rate (among those aged 16 and older), and the mean value of homes in the state. However, the addition of these covariates leads to greater attenuation in the estimated coefficients of the pupil-teacher ratio. We conclude that the effects of teacher wages are reasonably robust to other controls, whereas the pupil-teacher effects are more sensitive, and are likely overstated in the simplest models.

They also concentrate on the role of teacher wages specifically by looking at “cross-border” differences in schooling attainment, when states on each side of this border will have had different minimum teacher salary laws.  If this type of analysis to tease out causal effects sounds familiar it should – but this paper doesn’t seem to take a change in the minimum in two otherwise comparable county areas of different states, rather the fact that states have a varying minimum, so I don’t think we can really rule out that the measure could still be a proxy for other differences due to state policies [in other words it isn’t a difference-in-difference study, it is an IV model of differences between the country pairs].  They summarise:

In summary, our analysis shows that policies that increased teacher salaries substantially improved educational attainment among black children in the South in 1940. We
believe additional research will be useful to help fully develop our understanding of the mechanisms whereby these policies affected educational outcomes. One important possibility is that the higher salaries for black teachers in states with high minimum salaries led to a better-educated teacher workforce.

At face value this looks like a cool study that indicates higher teacher pay was associated with better educational outcomes, especially among disadvantaged groups.   Thoughts?