Teacher shortages and educational outcomes in developing countries: Empirical evidence from PISA-Thailand

Abstract Teacher shortages are among the most critical gaps that undermine educational performance, especially in developing countries, where there tends to be inequality in human resource allocation. This article therefore aims to study the impacts of teacher shortages on educational outcomes in a developing country by using Thailand as a case study. Using school-level survey data from the Program for International Student Assessment (PISA) in Thailand, estimates from the stochastic frontier analysis models indicate that teacher shortages are a serious problem that has a negative effect on the educational performance of Thai students, especially those living in rural areas. Furthermore, teacher shortages in certain subject fields are shown to have a spillover effect on other subject fields as well since teachers end up having to teach subjects that are not their specialization and to teach students of various ages and class levels mixed together. Therefore, effective teacher resource allocation should be tailor-made by taking into account the differences among schools such as school size or location since such factors play an important role in students’ educational outcomes in developing countries.


Introduction
Education is an important tool in building and developing knowledge, ability, and ethical behavior for individuals.And it also leads to sustainable social and economic development (Zidi et al., 2022).In terms of the economics of education, investment in education is more profitable than any other type of investment.It generates private benefit to students by providing them a better chance of being employed, a higher income from higher-paid jobs, and thus overall better social and economic status.Education also creates social benefits since countries with a well-educated and trained population are more likely to be economically successful and thus better able to provide good living conditions for their people (Ashenfelter et al., 1999;Card & Krueger, 1992;Heckman, 2012;Pholphirul, 2017;Rizzuto & Wachtel, 1980).
Elevating educational quality to a level that leads to national development depends on a plethora of factors.However, the factor that is considered the most important to achieving educational quality is "teacher quality."It is widely accepted that students who have been taught by highly qualified teachers have better educational outcomes than those whose teachers were of poorer quality (Hanushek, 1971;Murnane, 1975;Murnane & Phillips, 1981;Sanders & Rivers, 1996).
Teacher quality has long been acknowledged as crucial to effective educational input.It has also been hypothesized that teacher quality is a predictor of instructional quality, which in turn leads to better student achievement (Blömeke et al., 2016;Scheerens, 2016); Kyriakides et al. (2020); Kooli (2019) The Coleman report (1966) was one of the first studies conducted on the importance of teacher quality and how it affected students' learning.While it found that school and teacher quality were not statistically significant in determining learning outcomes and that factors related to other social and economic characteristics were more influential, later research did not agree and instead argued that teacher quality plays a vital role in students' learning.Given this role, improving teacher quality represents a crucial step in elevating a country's educational quality (Aaronson et al., 2007;Clotfelter et al., 2007;Hanushek et al., 2005;Kane et al., 2008;Rockoff, 2004) Burroughs et al. (2019) review the large body of literature on measures of teacher effectiveness, underscoring the diversity of methods by which the general construct of "teacher quality" has been explored, including teacher experience, education level, professional knowledge, and teaching effectiveness.Each of these concepts comprises a number of different dimensions and methods in measuring teacher quality.For example, a number of research studies have employed teacher experience as a variable to measure teacher quality, finding that students with teachers who had less than one year's teaching experience learned less than those with teachers with 10-15 years teaching experience (Murnane & Phillips, 1981;Hanushek et al., 2005;Rockoff, 2004;Kane et al., 2008). 1   In addition to teacher experience, teacher education is also found to affect student's educational outcomes.A teacher's highest educational degree obtained includes both the degree earned and the number of years teachers have spent studying their major.A review of the literature from Hanushek (1997) found approximately 40 research studies showing that teachers' highest education level was positively correlated with student outcomes.
Teacher knowledge and effective teaching methods are also included as key factors used to measure teacher quality.Baumert, et al. (2017), for example, investigated the significance of both subject knowledge and pedagogical expertise in generating high-quality instruction and student progress in secondary-level mathematics students in Germany.This research found that teachers' pedagogical expertise was distinguishable from their subject knowledge.
In addition to the factors cited above, another important factor significantly affecting teacher quality is inefficient teacher resource allocation (Sutcher et al., 2016).Allocation of teacher resources is found to be a common problem in most least-developed countries and many developing countries.Schools in remote areas in those countries, for example, do not have enough teachers to teach all subjects and for all grades.Thus, students from different grades have to be pooled in the same room so that the teacher can teach the same subject just once, at the same time to all students.Such a lack of teaching staff burdens teachers in such schools with a very heavy teaching workload as well as all the associated administrative work.Such large, mixed classes include students at different levels.In addition, teachers have to teach subjects that are not within their expertise.It thus comes as no surprise that such conditions should have negative impacts on student outcomes (Veeman, 1995;Veenman, 1996Veenman, , 1997) ) As a consequence, we aim here to quantify the effect of teacher shortages on students' educational outcomes in a developing country by using Thailand as a case study.Our research question is whether teacher shortages lower educational outcomes of Thai students.The next section analyzes the phenomenon of teacher shortages overall.Section 3 explains the context of teacher shortages in Thailand by using the PISA-Thailand dataset.Section 4 explains the methods employed in our study and Section 5 discusses whether and to what extent teacher shortages result in lower educational outcomes of Thai students by using PISA test score in three subjects which are Reading, Mathematics, and Science.Section 6 concludes and addresses teacher allocation, especially for schools in remote areas, and why it should therefore be considered in crafting policies to address educational inequality in Thailand.

Literature on teacher shortages
The teacher shortages arising in the world today are caused by various factors.From empirical studies conducted internationally, it can be concluded that the problem is brought on by two factors: population change and work environment (Podolsky et al., 2016).Population change has given rise to an increase in the elderly population.Since a significant segment of the teacher workforce comprises those who are nearing retirement age, when they retire the teaching workforce decreases.If this decline in the teaching workforce is not mitigated, a teacher shortage will arise.
As for work environment, this factor is considered to have the most significant influence on the teacher shortage in schools since these factors affect young people's career decisions-that is, whether they initially decide to become teachers in the first place and, if so, whether they decide to continue to teach vs. entering another profession.Work environment can be divided into five aspects: salaries and other compensation, teacher preparation, hiring and personnel management, induction and support for new teachers, and working conditions (Podolsky et al., 2016).
Salaries and other compensation are important factors in motivating people to work as teachers and in maintaining a robust teaching workforce.Salaries are also a vital factor influencing teacher allocation within different regions.Areas with high salaries and other compensation have an advantage in attracting teachers.Therefore, teacher shortages tend to be more severe in underprivileged schools located in remote rural areas since rural area schools tend to offer lower salaries and benefits than do schools located in urban areas, leading most teachers to seek employment in urban areas.
As for teacher preparation, teachers who are well prepared, both in terms of subject knowledge as well as classroom management, tend to be more self-confident with regard to educating their students as well as solving various everyday problems that arise in the school environment.Thus, such teachers are less likely to resign or quit in frustration than those without sufficient preparation (Smith & Ingersoll, 2004).
In terms of hiring and personnel management, delays sometimes occur during the recruitment process due to incorrect estimates of the number of teachers required and delays in the approval of hiring budgets by the central administration including asymmetric information between a vacant position in schools and requirements to work as a teacher.These factors make teacher shortages seem more severe than they actually are.
With reference to induction, orientation, and support for new teachers, schools that provide a good mentoring system to support new teachers-through coaching, monitoring, and assessing teaching performance-will most likely enjoy a higher teacher retention rate than those schools that do not have such a system in place (Kooli & Abadli, 2022;Smith & Ingersoll, 2004).
Finally, good working conditions cannot only attract teachers year after year but can also encourage teachers to not only remain in their teaching careers but to keep working in a particular school system.Good working conditions depend on four factors.The first factor is school leadership.If school administrators provide assistance in teaching and advice for problemsolving, teachers in such schools will be less likely to resign.The second factor is the opportunity for professional collaboration and shared decision-making.Schools with a good working team spirit that allows for collaboration in the workplace will be more likely to retain their teaching staff.The third factor is a system of accountability (Kooli, 2020).Schools that emphasize mere test taking and high test scores rather than focusing on students actually acquiring knowledge and that assess teaching performance based merely on students' test scores are, in effect, discouraging teacher retention.The last factor comprises resources for teaching and learning.Schools outfitted with materials/equipment for teaching and learning enhance teachers' morale by making it easier and more efficient for them to do their jobs, thus leading to greater teacher retention (Kooli & Abadli, 2022;Smith & Ingersoll, 2004).
Teacher shortages in certain areas can affect teacher quality in each school in that area, though perhaps not in an identical way.However, the result can be educational inequality even within a specific area.Teacher shortages tend to be more severe in underprivileged schools located in remote rural areas than in schools located in urban areas.If such rural schools do not have enough teachers, the teachers who are there commonly have to teach multigrade classrooms.This arrangement creates a greater teaching workload and means that teachers have less time to prepare lessons for each subject as well as less time to devote to each student, thus decreasing their teaching efficiency (Veeman, 1995;Veenman, 1996Veenman, , 1997) ) and leading to regression of the quality of education that each student should receive.Students taught in a multigrade classroom most likely have lower educational outcomes than students who are taught in a split class (Mariano & Kirby, 2009).Furthermore, when teachers who resigned from schools in rural areas had to be replaced, studies conducted abroad found that administrators were more likely to hire personnel without a degree in education or in a specific subject, without a teaching certificate, without any formal teaching preparation, or with less than five years teaching experience or none at all (Garcia & Weiss, 2019).Thus, accordingly, such practices lead to a deterioration of the quality of teaching and learning management and the overall quality of education, as reflected in students' educational outcomes.

Teacher shortages in Thailand
This section used school-level secondary data from the Programme for International Student Assessment (PISA) Thailand.The assessment is carried out every three years by the Organization for Economic Co-operation and Development (OECD) and surveys education systems in many countries to determine whether or not students are educated in terms of relevant preparedness for living and social participation in the future.Thailand is a country participating in this assessment 2 PISA-Thailand measures students' proficiency (literacy) in three foundational domains: reading literacy, mathematical literacy, and scientific literacy.Random sampling is conducted in two stages to get an accurate reflection of Thai students countrywide.The first stage samples stratified samples of schools across the country; the second stage samples 15-year-old students.And the 239 schools and 6,000 students involved are categorized according to their socioeconomic status. 3  The Institute for the Promotion of Teaching Science and Technology (IPST) collects data in each stage and assesses study results each year.Data collection in Thailand has been carried out every three years since 2000. 4  Analysis of data from PISA 2015 found that the survey data from 239 schools covered 88 percent of government schools (210 schools) and 12 percent of private schools (29 schools).Classified by school affiliation, it was found that 40 percent were schools affiliated with the Office of Basic Education Commission (formerly "general education") (OBEC 2) and that 16 percent were schools affiliated with the Office of Basic Education Commission (opportunity expansion schools) (OBEC 1). 5  The balance was made up of schools, institutes, and colleges, such as those affiliated with the Department of Education (Bangkok Education), the Office of Coordination and Development of Local Education, the Office of Vocational Education Commission (OVEC 2), and the Office of the Higher Education Commission (the Satit School).
Analysis of PISA survey data on teacher resources shows that on average government schools had about 78 teachers per school and private schools about 85 teachers per school.Most schools affiliated with OBEC 1 were small schools and had the fewest number of teachers, on average, 19 teachers per school, while private schools had as many as 106 teachers per school on average.Table 1 shows that most schools, almost 100 percent, employed teachers who had professional teaching certificates.Government schools had a higher proportion of teachers who had graduated with a bachelor's degree than did private schools.Schools affiliated with OBEC 2 (secondary education) had the highest proportion of teachers with a bachelor's degree (89.4 percent), followed by demonstration schools (78.3 percent), Office of Coordination and Development of Local Education (OCDLE) schools (69 percent), and Office of the Private Vocational Education Commission (OVEC 1) schools (68.5 percent).Most teachers in government schools (98 percent and above) were permanent teachers while private schools had the fewest number of in-service teachers (88.6 percent).There were only 48.1 percent of teachers who had not graduated from an education field working in private schools.
As for teacher shortages (displayed in Table 2), by comparing PISA-2009 and PISA-2015, it was found that even though the percentage of teacher shortage in school seems to be lower over time, many schools in Thailand still encountered these problems.Especially among small schools, there were shortages of reading teachers up to 47.2 percent and shortages of mathematics teachers up to 30.2 percent.Large and special large schools reported a relatively low percentage of teacher shortages in all subjects.Nearly 41-46 percent of special large schools reported no teacher shortage problem at all.
A key reason for teacher shortages in the Thai education system is the unequal allocation of teacher resources, which involves an area-based dimension.Most teachers prefer to work in large urban areas rather than in remote rural areas.This phenomenon is illustrated in Table 3, which indicates that in 2009 schools located in small villages encountered the greatest shortages of mathematics teachers (30.6 percent) and science teachers (22.5 percent) while most schools located in large urban areas (60 percent) did not encounter such shortages.
A study from Puncreobutr and Rattanatumma (2016) identified the reasons for the shortage of mathematics teachers at the Thai Basic Education level.Based on their qualitative results, they found that a shortage of mathematics teachers exists at both the primary and secondary levels.Furthermore, replacements were not made for retiring teachers, and capable people were moved to administrative jobs.They also pointed out that an underlying reason for the shortage of mathematics teachers was the fact that only a few students' showed interest in studying mathematics as their major.Besides, those teachers who had majored in mathematics may not choose a teaching career since teaching is considered a low-paying job.
In order to ameliorate the teacher shortage in mathematics and science, the Ministry of Education allows graduates with degrees in mathematics and science to work as teachers (most likely in government schools), not requiring them to have either a teaching certificate or a degree in education.OBEC 2 (secondary) schools had the greatest proportion of mathematics teachers with degrees in mathematics (89.4 percent), followed by demonstration schools (78.3 percent), and Office of Coordination and Development of Local Education schools (69 percent).In addition, Office of Vocational Education Commission schools also allow a high proportion of teachers with mathematics degrees to teach.
Management also plays an important role in supporting schools' teaching quality.This role extends from teacher employment to teacher dismissal to salary estimates and determination of teacher salaries and salary increases to determination of school budgets.The survey found that such authority differed for each type of school.For example, with regard to hiring, in general, school principals appeared to have the most authority to recruit and hire teachers.Private school principals had more authority in decision-making than did principals of other types of schools (93.8 percent of schools).Authorized persons at a national and regional level did not have decision-making authority in this regard while 94.4 percent of principals of Bangkok Education schools did have such authority.The main reason for the latter is that most (66.7 percent) Bangkok Education school budgets are determined by authorized persons at the national level.
Analyses of PISA test scores for each type of school (Table 4) indicates that test scores in each subject varied according to the levels of teacher shortages in each subject.Students studying in schools with large teacher shortages in reading, mathematics, and science got lower test scores in all subjects than did students in schools without such shortages.Teacher shortages were found especially in those small schools where teachers were assigned to teach students of different levels in a multilevel class, making it very difficult for students to learn according to a curriculum appropriate to their actual class level.Since the differences in educational outcomes are caused by a number of factors-school location, availability of teaching and learning materials/equipment, bureaucratic affiliation, and intra-school management-differences in test scores resulting from teacher shortages probably reflect some discrepancy.Thus, it is necessary to use an econometric instrument to control other factors as much as possible.This econometric model estimation is discussed in the next section.

Method
The analysis in this section uses school-level data to estimate the effects of teacher shortages on educational outcomes using the Stochastic Frontier Analysis (SFA) model, a model to estimate the Education Production Function. 6PISA test scores on all three subjects in each school, i.e. reading literacy, mathematics, and science, are used as an output from the education production function and can be divided into two factors, a teacher factor and a school factor, and written in a mathematical equation as Yi is the output of producer i, but in this case it is defined as school test scores, which determine the educational outcome in a school i. 7 Ti and Si are the teacher characteristics factor and the school factor input, respectively.It was supposed that the education production function equation determined the output as being the highest scores achieved (Production Frontier) under existing Ti and Si factors of production.It was additionally supposed that teacher quantity in different schools may have had different production efficiency.Education production efficiency (Eff i ) would determine how close school test scores were to the highest scores they could achieve, which can be defined as In order to estimate the stochastic production frontier model, it is necessary in the first place to specify f(i) further, which is normally assumed to take a translog form.The model can therefore be written as where ε i ¼ v i À u i , where v i is assumed to be i.i.d., symmetric and independent of u i .The error term of this equation, ε i ¼ v i À u i , is composed by a two-sided "noise" term and a nonnegative technical inefficiency term.Table 5 presents characteristics of independent and dependent variables used to estimate the above-mentioned equation. 8

Results and discussions
Table 6 presents estimation results using the Stochastic Frontier Analysis (SFA) model, in which the school factor was controlled.It was found that private schools had lower scores than did government schools (being the reference group) with statistical significance in all three subjects.Private schools financially supported by the government had lower scores than private schools that were not financially supported.Demonstration schools had the highest educational outcomes, higher than those of OBEC 1 schools (the reference group) by 15.8 percent for reading, 15.4 percent for mathematics, and 14.8 percent for science, all with statistical significance.OVEC 2 schools had the lowest educational outcomes, lower than those of OBEC 1 schools (the reference group) by 8 percent for reading, 10.2 percent for mathematics, and 9 percent for science, all with statistical significance.Similarly, Bangkok Education schools had outcomes lower than those of OBEC 1 schools (the reference group) by 7 percent for reading, 9 percent for mathematics, and 9.5 percent for science, all with statistical significance.This estimation shows that, compared to other variables, school affiliation had the greatest effect on educational outcomes.
As for school location, findings showed, with statistical significance, that schools located in a big city had higher PISA test scores than did schools located in urban and rural areas.Schools in the western region had the highest test scores, higher than schools located in the northeastern region (the reference group) by 3.8 percent for reading, 4.6 percent for mathematics, and 3.2 percent for science.School size did not have a statistically significant effect on educational outcomes.
The number of teachers with teaching certificates was another factor affecting educational outcomes.If schools had a 10 percent increase in teachers with teaching certificates, they saw test scores increase by 6-8 percent, with statistical significance.On the other hand, teacherrelated indicators such as student-teacher ratio or the number of teachers with a bachelor's degree were not found to have a statistically significant effect on educational outcomes.This is consistent with research studies conducted abroad that conclude that teachers' educational background does not have an effect on students' educational outcomes.
Teacher shortage analyses in which school characteristics variables are controlled in terms of school size, location, type (government school or private school), and affiliation reveal that if schools share such factors, those with teacher shortages tended to have lower educational outcomes than did schools without shortages.Estimation results reveal that schools with shortages of mathematics teachers tended to have statistically significant lower mathematics test scores by 3 percent while teacher shortages in other fields did not affect educational outcomes in those fields with statistical significance.
Tables 7-9 present estimations of three subjects (reading, mathematics, and science, respectively) classified by school affiliation.It can be seen that a Thai language (reading) teacher shortage decreases, with statistical significance, test scores (Table 7) and that a shortage of mathematics teachers (Table 8) significantly affects mathematics test scores of OPEC 2 schools.Schools with shortages of mathematics teachers had lower test scores by 24.8 percent while science teacher shortages had no significant effect on test scores of any school, no matter their affiliation (Table 9).Teacher shortages in a certain field had a negative effect not only on educational outcomes in a certain subject but on other subjects as well, as evidenced from mathematics teacher shortages in OPEC 2 schools, which had 18.7 percent lower reading scores (Table 7) and 20 percent lower science scores than did schools without mathematics teacher shortages (Table 9).

Conclusion and Policy Suggestions
Since teacher shortages affect the quality of teaching and thus educational outcomes in many developing countries, this study aims to analyze and estimate the effects of teacher shortages on students' educational outcomes by using Thailand as a case study of a developing country.Results from PISA-Thailand show that schools that encountered teacher shortages appeared to have slightly lower educational outcomes than did schools without teacher shortages.Estimation results indicate that while schools with mathematics teacher shortages had 3 percent statistically significant lower mathematics test scores, teacher shortages in other fields had no significant effect on educational outcomes in those fields.The estimation based on school affiliation found that Thai language teacher shortages significantly decreased test scores in demonstration schools and that mathematics teacher shortages significantly lowered mathematics test scores in OPEC 2 schools by 24.8 percent.On the other hand, science teacher shortages had no significant effect on test scores of schools no matter what their affiliation.
Teacher shortages in a certain field had a negative effect not only on educational outcomes in a certain subject but also had spill-over effects extending to other subjects as well, as evidenced by the fact that mathematics teacher shortages in OPEC 2 schools were associated with lower reading scores (by 18.7 percent) and lower science scores (by 20 percent) than test scores of those schools without math teacher shortages.Based on the study results, it can be seen that teacher shortages have a negative effect on students' educational outcomes in Thailand.This reflects the following problems in the teaching workforce market in Thailand and that is seen in many developing countries: (1) Problems related to inconsistency between teacher supply and demand.At present Thailand produces a large number of graduates in education while teacher positions are limited.This mismatch has been caused, at least in part, by downsizing in the government sector.In this regard, estimations of the number of teachers required to fill the teaching workforce should be more accurately made by determining the number of teachers needed in each school as well as their required fields.In addition, the fields for which each school requires teachers should be identified so that faculties of education can provide relevant teaching and learning consistent with such requirements.This should allow faculties of education to more accurately adjust their numbers of graduates to meet actual requirements.
(2) Problems related to inequality of teacher resource allocation.Large schools in urban areas have a considerable number of teachers and do not tend to encounter teacher shortages while small schools in rural areas commonly face such shortages.This means that students of different grade levels are often put together in a multigrade class and that teachers are expected to teach subjects they are not qualified to teach.To address this problem, more positions should be allocated to schools facing teacher shortages by offering additional remuneration to motivate teachers to teach in underserved schools in rural areas.Special hiring consideration should be given to teachers who are local people and who are local knowledge holders (such as local government officials, folk philosophers, local wisdom teachers, volunteer teachers, professional training students, teachers, or local people who are widely respected).
(3) Problems related to teacher personnel management.Schools (especially government schools) play a very small role in recruiting teachers.In addition, the recruitment process often uses examinations that do not reflect relevant standards and real-life teaching skills requirements.This is not to mention that the recruitment process often lacks transparency and that those who end up being hired are recruited as a result of knowing some local level education official (vs. on their merit as a potential teacher).Therefore, teacher recruitment needs a revamping, starting from teacher employment that should be offered based on one standard examination in each educational service area and giving schools themselves the authority to recruit teachers.After teacher hiring, classroom observation should be carried out rather than basing teacher assessment merely on examination scores.A guideline for teacher personnel management should be tailored to small schools facing resource shortages.The goal would be to improve the matchup between potential teachers and underserved schools.Incentives would include higher salaries and benefits, clear standards set for salary increases, and the addition of academic standing for organizing training that uses instructional media and having a broad ability to teach various subjects, etc.
A limitation of this study is that since frontier analysis is based on the production function measuring school performance, it requires mainly school-level data and omits using variables related to a student's individual characteristics such as gender, parent characteristics, learning behavior, and attitude toward school and education.In addition, since our school-level data is cross-sectional data, the primary limitation is that there is no evidence of a temporal relationship between teacher shortage and student's learning outcomes.Also, since PISA surveys do not collect longitudinal (panel) data, it is therefore not possible to establish a true cause and effect relationship between shortages and learning outcomes.

Table 1 . Average proportion of teachers classified by school types and affiliation (percent)
Source: Analyzed from PISA-Thailand in 2015.

Table 2 . Percent of school samples reporting teacher shortages classified by subjects and by school size, comparing 2009 and 2015 School size PISA-2015 No teacher shortage Slight shortage Moderate shortage Large shortage No teacher shortage Slight shortage Moderate shortage Large shortage
Source: Calculated from PISA-Thailand assessment results in 2003-2015 year round.

Table 4 . Teacher shortage and educational outcomes Variables Reading Mathematics Science Thai language teacher shortage
Source: Calculated from PISA assessment results in 2015 year round.

Table 7 . (Continued)
Standard Errors are in parenthesis.Control variables related to students' personal characteristics, students' family background, and school factor were included.

Table 8 . (Continued)
Standard Errors are in parenthesis.Control variables related to students' personal characteristics, students' family background, and school factor were included.