Tripping at the finish line: Misperceptions and secondary school completion in Argentina. Guest post by Carolina Lopez
This page in:
This is the 13th in this year’s series of posts by PhD students on the job market.
Capable students from low-income households can face hurdles that prevent them from completing their secondary schooling and obtaining a high school diploma. Small bureaucratic obstacles, that are easily negotiated by those with substantial parental support, may trip those students without such resources. And a lack of information about how inputs translate into outputs may prevent students from improving their educational achievement. Fryer (2011) and Fryer (2016) suggest that a potential explanation for students' failure to transform inputs into academic outputs could be a lack of adequate knowledge about the education production function, i.e., students do not know how to transform effort into improvements on academic achievement. In addition, it could be the case that students overestimate their own chances of graduation and consequently mis-calibrate the effort that they need to exert.
In my job market paper, I study whether providing information on how to get a high school diploma increases the likelihood of graduation among students in Argentina. I examine a particular hurdle that appears to be a significant barrier to high school graduation among students that have otherwise successfully completed the required coursework. The educational system allows a student who has failed some courses to continue studying during the next year, but at some point, she must pass those failed courses to graduate. This creates a scenario where most students can make it to the finish line, but still not get their diploma. This is a highly relevant issue. In this setting, secondary school enrollment is high (more than 90 percent of teenagers are enrolled in upper secondary school) but graduation rates are low (50 percent) even among students who attend until the last day of classes. The problem I posit is that some students are at high risk of not graduating based on their academic standing but unaware of this risk. In addition, based on my scoping field work, I also found that students have very little information about labor market returns to graduation which may influence their efforts to graduate.
For a subset of randomly selected public high schools, I provided information to senior students about the previous cohort's graduation rates based on academic standing at the beginning of the senior year. These statistics were meant to offer students the ability to map their own academic standing and chances of graduation. This treatment is called Production function. In another random group of schools, I provided information about employment rates and average earnings by level of education (the Returns to education treatment). And a third random group of schools served as the control group.
The two treatment interventions were conducted via a single visit to each school in the form of an information session with students. Of note, this is a very low cost treatment, arguably easy to scale-up. I combined a baseline survey and hard copies of individual academic records collected from each school to analyze the impacts of the intervention on graduation. The participants were almost 1,800 senior students, aged 18 on average, and 60 percent were female.
Evidence in economics and psychology shows that individuals tend to overestimate the probability of important outcomes leading to suboptimal decisions, especially for unskilled individuals (Choi et al, 2014). To analyze this concern, in the baseline survey I asked students for their perceptions of the likelihood that they will graduate on time. I compare this subjective measure with the estimated probability of graduation based on observable characteristics to create an indicator of confidence. I separate the students into two groups according to their academic standing at the beginning of the senior year: in good standing (those without failed subjects from previous years) and in bad standing (with failed subjects that need to be passed along with the subjects of the senior year). In Panel A and C of Figure 1, I show the distribution of the objective measure of confidence by treatment arm. Panels B and D show the difference between predicted graduation likelihood and self-reported perceived likelihood. Overconfidence is widespread for both groups of students. The concern is that those who over-estimate their likelihood of graduating at a given academic level may mistakenly underinvest in the actions needed to reach the finish line (to graduate).
Both treatments, on average, increase the probability of graduation. The Returns to education treatment increases the probability of graduation by 10 percentage points (almost 20 percent with respect to the control group); the Production function arm increases graduation by 5 percentage points (10 percent). The students with the greatest response in both treatment arms are those with the worst academic standing at the beginning of their senior year. Overconfident students who get the Returns to Education treatment experience a larger increase in graduation rates from the treatment than those who received the Production Function arm. Of concern with the Production Function arm for overconfident students was that providing this information to them might result in a discouragement effect – whereby the information would result in lower graduate rates — but this did not happen. Students who are underconfident (more likely to graduate based on academic standing than they realize) have increased graduate rates from both treatments. The model in my paper shows that under certain conditions, both under- and overconfident students increase their effort (and, hence, have higher graduate rates) when they receive information about the probabilities of graduation.
Mechanisms: is it perception or action, or both?
I test for two potential mechanisms for the treatment impacts found. First is that self-perception of graduation results in a change in actual graduation. Perceptions of graduation should only change if students updated their beliefs about the level of effort needed to obtain their diploma. This is only possible if they receive information about the actual probabilities, the effort that is required, and all the intermediate steps needed to successfully transform effort into graduation. I show that individuals who received the Production function treatment became more accurate with respect to their own chances of graduation.
The second mechanism relates to effort. I measured the role of effort devoted to the study time of students who are in bad standing at the beginning of the senior year. I focus on a dummy variable “passed at least one pending subject before the end of the academic year” (February 2020). Students who received the Returns to education treatment are 16 percentage points more likely to pass a pending subject (the mean of the control group is 28) and those who received the Production function arm are 6 percentage points more likely to pass a pending subject with respect to the control group, but in this case, the magnitude is nonsignificant at the standard levels. These results indicate that a single but targeted intervention for different types of students could help to ameliorate a detrimental cognitive bias.
Improving graduation rates, especially getting to the finish line for last-year secondary school students, can have life-long implications on earnings. This study explores a scalable, low-cost intervention, likely especially relevant for students who are disadvantaged and at risk of failing to complete high school because they lack the insights needed on the final steps needed to graduate and the relevance of graduation on future wages. The lack of such insights can result in less than the optimum level of effort. A modest intervention for senior students in Argentina can tackle the inaccurate beliefs students have about their future performance and the advantages of graduating to address the “graduation gap”, especially for poor-performing students.
Carolina Lopez is a PhD candidate at Brown University.
Join the Conversation