Student autonomy and academic performance

Sofoklis Goulas, Rigissa Megalokonomou 01 August 2021



During the COVID-19 pandemic, officials in many countries used school distancing to mitigate the spread of the virus, allowing for greater student autonomy (Burgess and Sievertsen 2020, Goulas and Megalokonomou 2020). Οfficials are interested in evaluating the consequences of the pandemic-induced student autonomy (OECD 2020, World Bank 2020). Understanding how high-performing students in particular respond to autonomy can provide focus for recovery strategies from pandemic-related learning losses and predict their success. 

In our recent work with Silvia Griselda (Goulas et al. 2021), we ask the following policy questions: What is the impact of compulsory class attendance on higher-performing students? Could autonomy improve higher-performing students’ outcomes? Because learning productivity may drop in academically diverse settings (Aucejo et al. 2018), we also ask: Is there heterogeneous demand for autonomy by classroom academic diversity?

First, we develop a simple model that provides insights into students’ time allocation problem. This model shows that students who maximise their utility with respect to in-classroom and outside-classroom learning, as well as leisure, may optimally choose to take a positive number of absences. This optimal number of absences may be positively correlated with student ability as the marginal productivity of out-of-classroom learning increases with ability. The optimal number of absences may also be negatively associated with classroom academic diversity to the extent that diversity translates to lower classroom learning productivity.

We test empirically the understanding developed in our theoretical framework. In particular, we estimate the effect of an increased autonomy policy for higher-performing students on short- and longer-term school outcomes. We exploit an institutional setting with high demand for autonomy in randomly formed classrooms. Identification comes from a natural experiment that allowed students with a prior-year grade point average (GPA) above 75% to miss 44% more classes without penalty. We combine attendance records, detailed transcript information and university admission data for students in two consecutive high school cohorts (one used as a control and one a treated cohort) in Greece. Using a difference-in-difference-in-differences approach, we find that allowing higher-achieving students to skip class more often improves their performance in high-stakes subjects and increases their university admission outcomes. Higher-achieving students in more academically diverse classrooms exerted more autonomy when allowed to.

For this study, we collected longitudinally linked transcript information from 107 schools covering more than 12,000 students in grades 10, 11, and 12 between the 2003/04 and the 2006/07 school years. The transcript data contain a student identifier, school year, grade, courses taken, scores for each course, the GPA, and the number of excused and unexcused absences (in periods). Absences can be excused only by a doctor or a guardian. Only whole days of absence can be excused.

Previous research has shown that flexible attendance requirements without sufficient time input autonomy has a negative impact on higher-performing students’ academic performance (Kapoor et al. 2021). Our paper moves beyond previous studies in two important ways. First, we identify the impact of time input autonomy – which allows for more self-study – on short- and longer-term outcomes of higher-performing students. Higher-performing students may have a propensity to better self-regulate, and therefore the ability to acquire more knowledge on their own (Zimmerman 2008). Second, we reveal the association between academic diversity of peers and demand for autonomy. In particular, we show that higher-performing students demand more autonomy when randomly assigned to a classroom with higher level of academic diversity.

An innovative autonomy policy

Near the end of the 2005/06 school year, the Greek Ministry of Education implemented a policy change intended to encourage students’ autonomy. The new policy provided eligible students with 50 additional excused class absences. Every student who had received a GPA higher than 75% the year before was eligible to take up more excused absences in the current year. The rationale for this policy change was that targeted students — those with a prior GPA above 75% — would have greater flexibility in making decisions related to their class attendance that best serve their own interests (i.e. time on self-study or leisure). We consider the cohort graduating high school in 2006 to be the control group and the cohort graduating in 2007 the treated group. The cohort graduating in 2007 was subject to the increased autonomy policy, introduced in 2006/07, in grade 12. Students could not manipulate in advance their eligibility for the increased autonomy policy, since the autonomy policy was unanticipated and eligibility depends on prior-year GPA.

Figure 1 plots the distributions of class absences in grade 12 of students before (cohort graduating in 2006) and after (cohort graduating in 2007) the increased autonomy policy, showing a shift of the distribution to the right when the autonomy policy was introduced.

Figure 1 Class absences distribution before (cohort 2006) and after (cohort 2007) the autonomy policy

How much autonomy was used?

Figure 2 shows the effect of the increased autonomy policy on class absences. Targeted students in the treated cohort took more excused class absences as a result of the increased autonomy policy. In particular, targeted students increased their excused absences by 0.13 standard deviations – roughly three additional classes relative to non-targeted students during the year the increased autonomy policy was in effect. Unexcused absences – the limit of which was not changed by the autonomy policy – remained statistically unaffected. 

Figure 2 The impact of the increased autonomy policy on class absences 

Impact of student autonomy

Figure 3 shows the effect of the increased autonomy policy on two types of subjects: high- and low-stakes subjects. We consider 12th-grade subjects that matter for high school graduation and for university admission and their 11th-grade equivalents to be high-stakes subjects. We consider 12th-grade subjects that matter for high school graduation but not for university admission and their 11th - grade equivalents to be low-stakes subjects. Targeted students’ performance in high-stakes subjects increased by 0.07 standard deviations due to the increased autonomy policy. Targeted students’ performance in low-stakes subjects and overall GPA remained unaffected by the increased autonomy policy. An investigation of longer-term outcomes shows that the university admission exam performance of targeted students improved as a result of the autonomy policy.

Figure 3 The impact of the increased autonomy policy on performance 

What drove the use of autonomy?

Figure 4 shows the differential impact of the increased autonomy policy on absences by classroom academic diversity. Academic diversity is proxied by the standard deviation of prior-year GPA in the classroom. Students are alphabetically assigned to classrooms at the beginning of high school and remain in their assigned classrooms for the preponderance of classes throughout high school. We find that targeted students quasi-randomly assigned to classrooms with higher academic diversity took more class absences when the autonomy policy was introduced. This suggests that learning productivity may be lower in academically heterogeneous contexts leading to more absences for higher-performing students in those settings. 

Figure 4 Differential impact of the autonomy policy on absences by classroom academic diversity 

The bottom line

As officials prepare policies to respond to the impact of pandemic-induced autonomy on student learning, there are lessons to be learned from policies that provided students with increased autonomy. Understanding how students can learn effectively away from school can widen the array of pandemic recovery policies that allow resources to be allocated where they are most needed and help policymakers predict their success. Our results demonstrate that autonomy in the form of relaxed school attendance for higher-performing 12th-graders may improve their performance, and that in the context of high-stakes exams, it may have significant long-term consequences on careers. 

More generally, the results highlight how compulsory class attendance can lead to allocative inefficiency. The allocation of student time to potentially less productive contexts (i.e., in academically diverse classrooms or lower-gravity classes) could result in poorer university and career placements and lower labour productivity.


Aucejo, E M, P Coate, J Fruehwirth, S Kelly, and Z Mozenter (2018), “Teacher effectiveness and classroom composition: Understanding match effects in the classroom”, CEPR Discussion Paper 13166.

Burgess, S and H Sievertsen (2020), “Schools, skills, and learning: The impact of COVID-19 on education”,, 1 April.

Goulas, S and R Megalokonomou (2020), “School attendance during a pandemic”,, 17 August.

Goulas, S, S Griselda, and R Megalokonomou (2021), “Compulsory Class Attendance versus Autonomy”, IZA Discussion Paper Number 14559. 

OECD (2020), “Education responses to COVID-19: Embracing digital learning and online collaboration”, Technical Report.

World Bank (2020), “How countries are using edtech (including online learning, radio, television, texting) to support access to remote learning during the COVID-19 pandemic”, Technical Report.

Zimmerman, B J (2008), “Investigating self-regulation and motivation: Historical background, methodological developments, and future prospects”, American Educational Research Journal 45(1): 166–183.



Topics:  Education

Tags:  student autonomy, school performance, Greece, student attendance

Senior Research Associate, Hoover Institution, Stanford University

Assistant Professor, School of Economics, University of Queensland


CEPR Policy Research