Education Quality across Indonesia's Districts Estimations from a Policy Experiment.

AuthorDharmawan, Goldy
  1. Introduction

    As a nation, Indonesia has always placed great attention on education. In the 1970s, the government embarked on one of the largest primary school construction programmes in the world (Duflo 2001). After primary school participation became universal in the mid-1980s, the government turned its attention to the junior secondary level (Suryadarma et al. 2006a). In addition to striving to increase educational attainment, the government also embarked on many programmes to improve the quality of education. Since the early 2000s, it has implemented several reforms related to teacher recruitment (Huang, Revina and Fillaili 2020), teacher professional development (Revina et al. 2020), teacher certification and remuneration (de Ree et al. 2018), school funding mechanisms (Majewski et al. 2013), and changes to the national examinations (Berkhout et al. 2020). In fact, since 2009, the government has consistently invested onefifth of its budget in the education sector (Kurniawati et al. 2018).

    Much has been written on Indonesia's education over the past two decades. (2) The consensus is that the country has witnessed magnificent gains in school participation, but the quality of education--measured by performance in standardized tests--continues to be low and stagnant. Beatty et al. (2021) echo findings from the latest international assessment, the Programme for International Student Assessment (PISA) 2018 (OECD 2019), that the quality of education has declined over the past two decades. For vocational and higher education, an appropriate measure of education quality is the graduates' labour market outcomes. Suharno, Pambudi and Harjanto (2020) find that vocational education does not provide students with relevant skills. And vocational schools are characterized by inadequate facilities and a lack of industry support. Therefore, graduates face difficulties in securing employment. The World Bank (2020) also finds similar issues with tertiary education. In summary, the notably higher public investment and many of the reforms we mention above have not led to higher quality education.

    From where should a country begin its mission to improve the quality of education? In this paper, we argue that one of the necessary inputs is information. If an education system intends to improve the level of learning that students accrue, then the information on learning outcomes must be available at a sufficiently disaggregated geographical level and with adequately high frequency to inform policymaking. Such information will provide an indication of the returns to public investments in education and enable policymakers to identify specific aspects in each geographical area that require attention. For a country that has decentralized its education service delivery, such information will enable the central government to direct support accordingly.

    Without such data, policy reforms or higher public investments in education are unlikely to be effective. Pritchett (2015) states that, without useful information, accountability relationships among actors in the education system break down. Empirically, Chun and Gentile (2020) find in their benchmarking exercise that providing the public with information on learning gaps and the availability of information that allows real-time data-driven decision-making are two out of six crucial indicators correlated with higher learning outcomes.

    At present, such information on learning levels is not readily available in Indonesia. (3) National examinations are only conducted in grades nine and twelve, and being high-stakes summative examinations, they have been prone to systematic cheating (Berkhout et al. 2020). While the government has successfully reduced cheating in national examinations, information on learning outcomes is needed at earlier education levels. Similarly, international assessment results such as PISA and TIMSS (Trends in International Mathematics and Science Study) are only available at the national level.

    In this paper, we use a policy-induced natural experiment to estimate learning levels at the district level. The national examination in grade nine is our main data source. As mentioned above, national examinations have been compromised by systematic cheating. Therefore, to estimate counterfactual examination scores (had there been no cheating), we base our calculations on the computer-based testing (CBT) policy. Specifically, we use the variation in CBT policy implementation within provinces to predict the national examination performance of schools that were still using paper-based testing. We implement the procedure for 2017 and 2018. To our knowledge, this is the first district-level measurement of education quality for Indonesia.

    Our results indicate that the quality of education varies significantly across districts. The gap between the lowest-scoring and the highest-scoring districts indicates that, in the lowest quality districts, hardly any learning is taking place despite students being enrolled in school for nine years. Similarly, we find that, within a district, the average gap in the level of learning between students enrolled in a low-scoring school and those enrolled in a high-scoring school is as much as six years of education.

    We organize the rest of the paper as follows. The next section provides an overview of some of the trends in educational attainment and education quality since 2000. The third section describes the CBT policy, estimation strategy and data source. The following section provides the results, and the fifth and final section concludes.

  2. Education Outcomes since Decentralization

    Indonesia's decentralization policy became effective in January 2001. From then on, the responsibility of delivering education services was transferred to the country's subnational governments. Provincial governments oversee senior secondary education, while district (or local) governments manage basic education - comprising primary and junior secondary education.

    In this section, we analyse trends in educational attainment and education quality using the Indonesia Family Life Survey (IFLS). The IFLS is a household panel data set that is representative of 80 per cent of the Indonesian population. (4) It was first carried out in 1993, with four further rounds implemented between 1997 and 2014. In comparison to similar datasets in other developing countries, IFLS has very low attrition rates. Strauss et al. (2016) state that 92 per cent of respondent households in the first round were able to be re-interviewed in all subsequent four rounds.

    The main reason for using IFLS, as opposed to other nationally representative household surveys, is the numeracy skills module that covers respondents between the ages of seven and twenty-four. Since 2000, the IFLS has been collecting a consistent measure of basic numeracy skills covering the first five grades in the Indonesian curriculum. The data allow us to measure trends in basic mathematics skills over a long period of time for a large part of the population. To our knowledge, no other publicly available data set in Indonesia contains comparable measures.

    Figure 1 shows the net enrolment rates in primary, junior secondary and senior secondary levels just before decentralization until the latest IFLS wave. Primary school enrolment became universal in 1984 (Suryadarma et al. 2006a), and Indonesia successfully maintained this level until 2014. Between 2000 and 2014, junior secondary and senior secondary net enrolment rates increased. The enrolment rate for junior secondary schools was 71 per cent in 2000 and increased to 89 per cent in 2014. At the senior secondary level, the enrolment rate was 47 per cent in 2000, which increased to 71 per cent in 2014.

    The improvements in overall net enrolment rates were accompanied by narrowing gaps between male and female students. Figure 2 compares the probability of enrolment by age and gender. In 2000, we observed that boys had, on average, a 1.6 percentage point higher probability of enrolling in school than girls. And this gap widened with age, indicating that boys had a higher probability of studying to a higher level, while girls stopped studying at an earlier age. However, the gender gap had disappeared by 2014, indicating that both boys and girls had equal opportunities to attain education to a higher level.

    Figure 3 compares the probability of being enrolled in school by island of residence, namely Java/Bali versus off Java/Bali. In 2000, the enrolment probabilities were very similar between the two groups, except for a significantly higher but small probability of being in school for the sample older than eighteen years old in Java. By 2014, the gap had disappeared as residents outside Java/Bali caught up.

    In Figure 4, we compare the probability of being in school by age for children from the bottom 20 per cent, middle 20 per cent and top 20 per cent of household consumption. In 2000, we observed a gap of almost 15 percentage...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT