It has become almost cliché in certain policy circles—notably in Latin American countries—that improving education quality requires increasing teacher pay.
How did this idea become the conventional wisdom?
It all begins by recognizing that good teachers are far more important for student learning than, say, how beautiful a school is or how often standardized tests are administered to children. There is robust, credible evidence in support of this basic premise.
Beyond this basic premise, things get murkier. Wages, the argument goes, are just like any other market price: the laws of supply and demand govern them. Because of that, characteristics of non-teaching jobs available to potential teachers—primarily pay—affect who chooses to enter the teaching profession.
The key prediction that follows from this argument is that when pay in other occupations is relatively more attractive than pay in teaching, potential teachers may prefer to work in those occupations instead of becoming teachers. Teacher pay relative to other occupations may, therefore, affect education quality to the extent that it affects the decision to become a teacher for many talented, motivated individuals who would presumably make great teachers.
Under the presumption that relative teacher pay affects education quality downstream, many studies in Latin America and globally have tried to establish whether teachers earn less or more that comparable professionals in other occupations. Most studies reach the same three conclusions:
- Teachers are underpaid relative to other occupations.
- Teachers work fewer hours/days than workers in other occupations.
- Pay per hour/day is actually higher among teachers than workers in other occupations.
And thus, these three findings have become the conventional wisdom about teacher pay.
The problem with the conventional wisdom in this case is that it is based on faulty data, drawing mainly from household surveys that are not well suited for the task at hand. For instance, household surveys rarely collect information about teachers’ actual outside labor market options, their teaching ability, or the fact that teachers may possess other (unobserved) attributes such as motivation or public-service inclination that are related to compensation in the labor market. Put simply, the data limitations in existing studies are simply too onerous to draw credible conclusions about teacher pay policy.
In a recent study sponsored by the Inter-American Development Bank, we aim to overcome the shortcomings that pervade much of what (we think) we know about teacher pay. We assembled a novel and unique dataset from individual-level confidential, restricted access sources for all college graduates and for the subset of them that apply for public teaching jobs in the country of Colombia. These data include individuals’ educational histories, test-based measures of teaching ability and labor market trajectories. We ask whether the three widely accepted conclusions about teacher pay hold when they are subjected to these improved data and a more rigorous methodological standard—one that truly aims to identify causal attribution as opposed to merely establish statistical association.
We found some surprising results.
First, new teachers in Colombia actually earn more in their first three years of teaching than the average college graduate in the country. Moreover, once we account for public-service inclination and teaching ability, the difference between the wages of teachers at the beginning of their careers and those of their comparable counterparts who work in other occupations is even greater. These findings cast doubt on the widely held belief that teachers are underpaid relative to other occupations that also require a college degree.
Second, rather than working less, teachers actually work more days per year than their comparable counterparts in other occupations—possibly as a result of more stable working conditions. This additional teacher work effort throughout the year partially explains why take-home pay is greater for teachers than non-teachers, at least in the first few years on the job.
Third, once you account for greater take-home pay and greater work effort, pay per day is greater for teachers than for comparable non-teachers. This finding corroborates earlier evidence that pay per unit of time may in fact be greater for teachers than for comparable workers in other occupations.
While our findings challenge some—two out of three, to be specific—beliefs in the conventional wisdom about teacher pay, a few caveats are in order. One is that we only observe labor market trajectories of teachers and other workers early in their careers—in their first three years, to be precise. It may well be the case that earnings of workers in other occupations overtake those of teachers later on, especially if the earnings of those workers rise faster with experience than the earnings of teachers.
Another caveat is that our findings in the context of Colombia may not generalize to other regions or countries. Latin American countries share many of Colombia’s labor market features. More developed nations may not. For example, formal labor market opportunities in Colombia are not as abundant as they are in richer nations such as the U.S., Canada or other high-income countries. In such settings, teaching may not be as attractive a career option as other occupations.
On the whole, our results call into question policy initiatives that point to low teachers salaries as the main reason why teaching cannot attract highly qualified college graduates to the profession. These initiatives assume that raising salaries will—by itself—automatically improve the pool of potential teachers. Other studies have shown, for instance, disappointing short-term results of large, unconditional teacher pay increases. The fact that early-career teachers earn more than similarly qualified professionals in other occupations suggests that simply raising salaries may not be enough to generate a stronger teaching workforce. Alas, improving education quality may actually require a more nuanced and holistic strategy.