Exploring Differences in the Distribution of Teacher Qualifications across Mexico and South Korea: Evidence from the Teaching and Learning International Survey
by Thomas F. Luschei, Amita Chudgar & W. Joshua Rew — 2013
Background/Context: Although substantial evidence from the United States indicates that more qualified teachers are disproportionately concentrated among academically and economically advantaged children, little cross-national research has examined the distribution of teacher qualifications across schools and students. As a result, we know little about how different institutional contexts, policies, and priorities influence children’s access to qualified teachers.
Research Question: Our research questions are: (1) Are the qualifications of lower-secondary teachers within and across Mexico and South Korea distributed uniformly across schools? (2) If not, does the distribution of teacher qualifications in each country favor less or more advantaged children? (3) How can dissimilarities in teacher-related policies and educational priorities help to explain differences in the distribution of teacher qualifications across Mexico and South Korea?
Research Design: We employ secondary analysis of data from the Teaching and Learning International Survey, which was conducted by the Organization for Economic Cooperation and Development in 2007/2008. We use these data to describe the distribution of various teacher qualifications across communities of different sizes and across schools with varying levels of parental education. We also explore cross-national differences in institutional priorities and teacher-related policies. We compare Mexico and South Korea because while these two countries are similar in the level of teacher hiring and assignment, they are quite different in terms of their general commitment to educational equity.
Findings/Results: We find that the distribution of qualified teachers in South Korea is skewed toward disadvantaged children, while Mexican teachers tend to be distributed in a way that favors more advantaged students. Specifically, in South Korea students living in rural areas and those in schools with lower average parental education have greater access to better educated and more experienced teachers. The opposite occurs in Mexico.
Conclusions/Recommendations: We argue that these differences are due to both explicit policies and a greater commitment to educational equity in South Korea, relative to Mexico. Moreover, these differences are likely to be related to large cross-national differences in educational performance and equity.
Increasing evidence from the United States suggests that consistent access to high-quality teachers plays an important role in raising student achievement and narrowing achievement gaps stemming from differences in students socioeconomic status, or SES (Aaronson, Barrow, & Sander, 2007; Hanushek & Rivkin, 2004; Nye, Konstantopoulos, & Hedges, 2004). At the same time, research indicates that high-quality teachers in the United States often sort into schools and communities in a way that skews the distribution of teacher quality toward more advantaged and higher-achieving children (Bacolod, 2007; Clotfelter, Ladd, & Vigdor, 2005, 2006; Lankford, Loeb, & Wyckoff, 2002). Although little international research has explored the phenomenon of teacher sorting, it is likely that similar sorting patterns exist in other countries, and that these patterns are strongly related to educational equity and quality. Cross-country differences provide an important source of variation to examine the relationship between teacher policy and educational outcomes (Tatto, 2008). In the case of teacher sorting, a cross-national perspective can allow us to observe variations across countries in the extent of teacher sorting patterns that may in turn be related to policy measures to ensure equity. Identifying such variation and understanding its sources may also contain important lessons for the United States. Yet despite the clear equity implications of teacher sorting, little research has systematically compared patterns in the distribution of teachers within and across countries, and few researchers have attempted to relate differences in these patterns to differences in policies and institutional arrangements.
In this paper, we use data from a comprehensive cross-national survey of teachers and school leaders, the Teaching and Learning International Survey (TALIS), to examine differences in the distribution of teachers in Mexico and South Korea. The choice of these countries is based on similarities in the level of teacher hiring and placement, which in both countries is relatively centralized compared to many other participating countries. This similarity means that more qualified teachers cannot bargain locally with schools and districts to secure the most desirable working conditions, which could lead to sorting of qualified teachers. Instead, more central actors make decisions regarding which teachers teach where, and can, in theory, ensure more equitable patterns of teacher quality. At the same time, educational priorities and teacher-related policies in these countries are quite different. As a result, we can examine how variation in policies and priorities relate to differences in the distribution of teachers across schools. Our specific research questions are: (1) Are the qualifications of lower-secondary teachers within and across Mexico and South Korea distributed uniformly across schools? (2) If not, does the distribution of teacher qualifications in each country favor less or more advantaged children? For example, are more qualified teachers more likely to work in urban versus rural settings, or in schools with higher average parental education? (3) Assuming differences in the distribution of teacher qualifications across Mexico and South Korea, how can dissimilarities in teacher-related policies and educational priorities help to explain these differences?
THE DISTRIBUTION OF TEACHERS AND RELATED POLICIES
Increasing evidence from the United States demonstrates that teachers of disadvantaged and minority children often have lower educational attainment, experience, or ability than teachers of more advantaged children (Bacolod, 2007; Clotfelter et al., 2005, 2006; Jackson, 2009; Lankford et al., 2002). For example, Lankford et al.s (2002) study of New York State found that highly qualified teachers sort into schools with students who are more economically advantaged and have higher levels of academic achievement, while disadvantaged and minority students are more likely to have less skilled teachers. Hill (2007) found that although teachers with higher content-specific knowledge of mathematics are on average more effective in raising students mathematics test scores, more knowledgeable teachers tend to be disproportionately concentrated in schools with more economically advantaged children.
While growing evidence makes clear that teachers in the United States are not evenly distributed across students, there is less evidence regarding the policy levers or institutional settings that influence the distribution of teachers. Boyd, Lankford, Loeb, & Wyckoff (2003) argue that teacher sorting in the United States is due to a combination of teacher labor market features, including district-based single salary scales, seniority-based recruiting, and reliance on local property taxes for funding. In the first case, uniform salaries across schools within districts make it impossible for difficult-to-staff schools to raise salaries without comparable increases for teachers in other schools in the same district. Without salary incentives to teach in difficult schools, more qualified teachers are more likely to choose schools with more pleasant working conditions. Second, under seniority-based recruiting, veteran teachers are often recruited away from difficult-to-staff schools once they gain experience, thereby leaving more difficult schools with less experienced teachers. Finally, because school districts rely in large part on local property taxes, wealthier districts are often able to pay higher salaries to recruit more qualified teachers.
Research in the United States has suggested a few potential policy levers to increase equity in the distribution of teachers, but researchers generally identify large gaps in our understanding of how to remedy teacher quality inequities. Lankford et al. (2002) discuss the possibility of salary incentives and signing bonuses targeting difficult-to-staff schools. The authors also emphasize that understanding and addressing teachers preferences for both salary and working conditions are key to closing teacher quality gaps. Although substantial evidence indicates that teachers respond to salary differences when deciding where to teach, Lankford et al. (2002) argue that working conditions, such as school resources and facilities, class size, and student characteristics, are also important in influencing teachers decisions about where they work. In a study of New York State teachers, Boyd, Lankford, Loeb, & Wyckoff (2005) found that one key determinant of teachers location decisions is proximity to home: the authors found that an individual is twice as likely to teach in a region within five miles of his or her hometown as in a region that is 20 miles away. This sensitivity to distance suggests that teachers are very committed to teaching (and not just living) in their hometowns, which in turn suggests that many teachers prefer to teach students similar to themselves.
In a national study using data from the Baccalaureate and Beyond Longitudinal Study, Bacolod (2007) found that working conditions are more important than salaries in determining where new teachers work. Moreover, the author found that teachers with higher SAT scores and college grade point averages are more likely to teach in suburban schools than in central city schools. Although the author acknowledges the potential impact of district hiring policies on teacher sorting, she did not examine their role in explaining these patterns. As Lankford et al. (2002) point out, many questions remain regarding appropriate policies to address teacher quality differences across schools.
One challenge to conducting analyses of teacher sorting is that relevant policies, practices, and institutions often vary little within a single district, state, or country. For example, teacher distribution research in the United States assumes an environment in which teachers negotiate with and are hired by local school districts or schools. Within the context of teacher labor markets, this research concentrates on supply-side concerns related to teachers availability, bargaining, and willingness to work in certain settings. As a result, this research often highlights the importance of teachers preferences in determining their working locations. In other countries, teacher hiring is much more centralized and can occur at the regional or national level (OECD, 2005, 2009). As a result, demand-side issues become more salient in the discussion of how to ensure an equitable distribution of teachers. Emerging evidence suggests that teacher policy differences between the United States and other countries may be related to cross-national differences in patterns of teacher quality. For example, in an analysis of eighth grade data from the 2003 Trends in International Mathematics and Science Study (TIMSS), Akiba, LeTendre, and Scribner (2007) found large cross-country differences in the access of economically disadvantaged students to qualified teachers. Whereas many countries have teacher quality opportunity gaps that place lower-income students at a disadvantage, the distribution of teachers in some countries, like Japan and South Korea, is much more equitable.
Akiba et al. (2007) suggest that cross-country differences in teacher qualification gaps may be due in part to the level of centralization of the education system. That is, more decentralized systems may have larger gaps in teacher quality across schools due to the ability of qualified teachers to negotiate and choose schools with more desirable working conditions. In contrast, centralized systems are more likely to assign teachers to schools where they are needed, regardless of teachers preferences. The authors give the example of Japan, in which teachers are hired regionally rather than locally, and then periodically rotated across schools to reduce school-based gaps in teacher quality. The authors also found that in South Korea, more qualified teachers are disproportionately concentrated among disadvantaged children (Akiba et al., 2007).1 Kang and Hong (2008) argue that this negative opportunity gap in South Korea is due to uniformly high-quality teachers, periodic rotation of teachers to new schools, and incentives to teach disadvantaged children.
THE MYSTERY OF TEACHER QUALITY
The first step in assessing whether disadvantaged children have equal access to qualified teachers is to identify what makes for a qualified teacher. Volumes of educational research have addressed this question, but little consensus has emerged (e.g., Wayne & Youngs, 2003). Goldhaber (2002) has referred to the problem of identifying the characteristics of good teachers as the mystery of teacher quality. Evidence of the importance of teachers educational attainment and experience beyond three to five years is inconclusive, despite the importance of these two characteristics in teacher compensation systems (Hanushek 1986, 1997; Ferguson, 1991; Ferguson & Ladd, 1996; Greenwald, Hedges, & Laine, 1996; Rivkin, Hanushek, & Kain, 2005). There is some evidence that teachers test scores and other measures of teacher ability or knowledge are positively related to student achievement (Ferguson, 1991; Ferguson & Ladd, 1996; Goldhaber & Brewer, 2000; Monk, 1994; Monk & King, 1994; Tatto, 2008). In particular, teachers knowledge of the subject for teaching appears to positively contribute to students academic achievement (Hill, 2007; Hill, Rowan, & Ball, 2005). Over time, improvements in data quality and analytical sophistication have shed increasing light on the importance of various teacher characteristics. Studies using longitudinal data controlling for prior student achievement to account for the non-random selection of teachers and students into classrooms and schools have resulted in new and important conclusions regarding teacher quality. For example, Clotfelter et al. (2007) found that teachers experience, test scores, and licensure positively affect student achievement, particularly in mathematics.
Studies exploring the distribution of qualified teachers have used a multitude of teacher quality measures, including teachers SAT scores and college grade point average (Bacolod, 2007), selectivity of teachers undergraduate college, scores on certification exams, experience, education level, and certification status (Lankford et al., 2002); certification, education level, undergraduate major, and experience (Akiba et al., 2007); education, experience, and self-reported readiness to teach (Luschei & Chudgar, 2011); and education and experience (Luschei & Carnoy, 2010). Because most of these studies employ secondary data analysis, they are constrained by the data available to them. Limited to available teacher quality measures, researchers cannot hope to adequately capture the complexity of teaching and teacher quality.2 On the other hand, these measures are, for the most part, observable by educators and policy makers, and in the case of education and experience, they determine teachers salaries across most of the globe (OECD, 2005). Consequently, the distribution of these teacher characteristics can illustrate the priorities and commitments of education systems to ensure equitable access to educational resources for all children.
BACKGROUND: MEXICO AND SOUTH KOREA
Mexico and South Korea make for a good comparative case because they are similar in terms of level of centralization but quite different across various measures of social, economic, and educational equality. The two countries also exemplify very different national education strategies, as we discuss below. Both countries are also non-European members of the Organization for Economic Cooperation and Development (OECD). The OECD contains 34 member countries, most of them in Europe. Generally, membership in the OECD implies high levels of national development and income, a commitment to democratic values, and relatively high levels of human development (OECD, 2011). Mexico and South Korea both participated in the OECDs 2009 Programme for International Student Assessment (PISA), as well as the OECDs TALIS study. Consequently, we are able to draw on multiple sources of valid and recent data to describe differences in their economic and educational contexts. Below we provide a brief background of these contexts in both countries.
Economic and Educational Context
South Korea and Mexico vary considerably in social, economic, and educational contexts (Table 1). Although Mexico had a higher per capita gross domestic product than South Korea in 1960 ($1,600 compared to $1,300), the two countries have experienced a reversal of economic fortunes in the past 50 years (Hansen, 2006, p. 627). By 2009, South Koreas gross domestic product was more than double that of Mexico (World Bank, 2011). In both countries, financial commitments to education grew considerably during the same period. In 1954, South Korea spent only 0.1% of its national income on public education, compared to Mexicos 1.0%. By 2002, South Koreas spending on public education represented 4.8% of its national income, compared to 5.1% in Mexico (Hanson, 2006).3
Hanson (2006) argues that reversals of economic and educational fortunes in the two countries stem largely from differences in the educational commitments and strategies of their national governments, noting that nations make choices as to how to invest their scarce resources, and those choices represent expressions of national priorities (p. 640). South Koreas priority after the conclusion of its debilitating civil war in 1953 was to ensure high literacy levels and a strong basic education system, in order to reach its entire population. In fact, South Koreas spending on public education as a percentage of GNP increased from 0.1% to 4% of national income between 1954 and 1960 alone. As Hanson argues, South Koreas significant investment in education created a virtuous cycle, in which investments in education contributed to rapid economic growth, thereby providing additional revenues to invest in education.
Mexicos substantial educational commitments have not resulted in similar gains in attainment and achievement, especially at the secondary level. South Korea and Mexico had similar net primary enrollment rates (99% and 98% respectively) in 2008, but attainment and achievement diverge considerably at the secondary level. In 2008, secondary net enrollment in South Korea was 95%, compared to 72% in Mexico (World Bank, 2011). While South Koreas upper secondary graduation rate of 93% was among the highest of all OECD countries in 2008, Mexicos upper secondary graduation rate of 44% was above only Turkey among OECD countries (OECD, 2010a). Achievement outcomes among 15-year-old students also differ considerably between the two countries. In the 2009 PISA, South Korea scored among the top 5 of 65 participating countries in both reading and math, whereas Mexico scored in the bottom 18 countries in reading and the bottom 15 in math (OECD, 2010b).
Differences in revenues and educational spending targets may help to explain some of the differences in educational outcomes across South Korea and Mexico. As noted above, the Mexican government spent a greater percentage of national income on public education than South Korea in 2002. Mexico also devoted a higher percentage of government expenditures to education than South Korea in 2002 (24% compared to 17%); internationally, Mexicos relative investment in education was among the highest in the world. However, Mexicos government collects much less tax revenue than South Korea, meaning that its overall government expenditures are much lower on a per capita basis. Mexico also has a much greater percentage of school-age children than South Korea, so educational resources must be spread among a larger population (Hanson, 2006).
Disparate educational outcomes also stem from differences in the priorities and decisions of national governments. For example, Mexico places a much greater emphasis on tertiary education relative to South Korea. In 2007, the Mexican government spent more than three times (3.3) as much per student in tertiary education relative to primary education, whereas South Koreas tertiary to primary spending ratio was 1.6, or half that of Mexico (OECD, 2010a, Table B1.1a). Because more advantaged students are more likely to attend tertiary education, a greater financial commitment to this level of education suggests a stronger government commitment to wealthier citizens.
CONCEPTUAL FRAMEWORK, DATA, AND METHODS
Two key considerations in comparing the distribution of teachers cross-nationally are the level of educational decision-making (central versus local) and the overall commitment to equity within the education system. In more centralized systems, decision makers have a greater opportunity to hire and place teachers in a way that provides equal access of children to qualified teachers, or that concentrates more qualified teachers with needier children. In decentralized systems like the United States, school leaders are more likely to negotiate directly with teachers. Leaders in schools with higher salaries and more desirable working conditions will have an advantage in attracting more qualified teachers, whereas leaders in more challenging or lower-resource schools have less power to recruit and hire good teachers. Consequently, more centralized systems are more likely to have a uniform distribution of teachers, or a distribution that favors less advantaged children, all else equal.
The overall culture of equity within a school system is also likely to influence the distribution of teachers and could interact with the level of educational decentralization in determining patterns of school and teacher resources across schools. For centralization to have an equalizing effect on school resources, education officials must explicitly seek to provide equitable access for all children to school and teacher resources. In a system that is committed to equality of opportunity, decision makers are more likely to implement policies that ensure an equitable distribution of teachers, such as incentives for teachers to work with disadvantaged children or the periodic rotation of teachers. In contrast, if actors in a centralized system are more interested in favoring the status quo or assuaging political clientele, centralized decision-making can result in a distribution of teachers that is skewed toward the children of wealthier or more politically powerful citizens. For example, in an analysis of sixth grade teachers in Uruguaywhich has a much more centralized system than the United StatesLuschei and Carnoy (2010) found that the distribution of teachers systematically favors high-SES children. The authors suggest that this pattern results in part from an institutional climate that favors more advantaged children, coupled with a teacher concurso, or competition, that gives first choice of schools to the highest-scoring teacher candidates.
In both South Korea and Mexico, teacher hiring is beyond the control of local schools, thereby diminishing the influence of school-based bargaining and negotiations. In effect, by selecting these two countries, we control for the level of centralization and concentrate on differences in institutional and educational environments. By most available measures, South Korea exemplifies a much stronger culture of equity and a more explicit focus on equalizing teacher resources across students than does Mexico. As a result, we hypothesize that the distribution of teachers in South Korea will be much more equitable, or even skewed toward more disadvantaged children, relative to Mexico.
DATA AND VARIABLES
To explore our primary research questions, we use data from the 2007/2008 application of TALIS. Conducted by the OECD in 24 countries, TALIS sampled 200 lower-secondary schools in each country and 20 teachers in each school. Questionnaires were also applied to school leaders. The resulting data provide extensive, rich, and nationally representative information related to teachers characteristics, working conditions, preparation, beliefs, attitudes, and practices across a diverse group of countries (OECD, 2009).
We rely primarily on data from TALIS questionnaires of lower-secondary teachers. The South Korean data include 2,970 teachers; 3,368 teachers participated in Mexico. In each country we examine a set of teacher- and school-level variables (Table 2). Teacher variables include gender, education level, experience, whether a teacher has received training in the subject he or she teaches, and a self-reported measure of efficacy. The self-efficacy measure is a composite of four items on the teacher questionnaire that measure teachers self-reported success in educating their students. We examine the distribution of these teacher characteristics across school characteristics that are based on previous research regarding the characteristics of schools and communities that may influence teachers preferences about where they teach. Specifically, we examine differences across various community sizes (village, small town, town, city, large city) and different school-level averages of parental education. We also correlate school-level characteristics of teacher characteristics with measures of working conditions in the school (material shortage, student delinquency, and teacher morale) as discussed below.
In all sorting analyses, we limit the data to include only public schools. This is because private schools follow different rules and operate under different levels of autonomy in hiring teachers. For example, measures of school autonomy available in the TALIS data indicate that, in both Mexico and South Korea, autonomy over hiring teachers is substantially greater in private schools. In Mexico, private school principals are 70% more likely to report considerable autonomy in selecting teachers for hire, and the school autonomy index is 2.84 points higher in private schools. In South Korea, the respective measures are 23% and 1.30 points higher in private schools relative to public schools. The international standard deviation for the school autonomy index is 1, so these differences are substantial.
Our analysis begins with descriptive statistics of the major teacher and school variables mentioned above. In all analyses, we use sample weights to ensure that results are nationally representative. Below we describe the specific methods to approach each of the primary research questions.
Research Question 1: Are the Qualifications of Lower-Secondary Teachers within and across Mexico and South Korea Distributed Uniformly across Achools?
To describe the distribution of teacher qualifications in each country, we report percentiles of a set of teacher characteristics across public schools, following methodology used by Lankford et al. (2002). If teachers are distributed uniformly, then teacher characteristics should not vary noticeably across schools. To explore whether this is the case, we calculate school averages of 7 teacher characteristics to generate 7 school-level measures: percentage of male teachers, percentage of teachers with less than bachelors degree, percentage of teachers with more than bachelors degree, percentage of teachers with 0-2 years experience, percentage of teachers with more than 20 years of experience, and average teacher self-efficacy. Next, we order each of these school-level measures from the lowest to highest value on the measure. For instance, we list the school-level percentage of teachers with more than a bachelors degree from lowest to highest for all 200 sample schools in Mexico. Next, we examine the bottom, middle, and top of this distribution, or the schools that fall at the 10th, 50th, and 90th percentile of the distribution. If the distribution of these characteristics is uniform, on average, schools at the 10th percentile should not look very different from schools at the median or schools at the 90th percentile. If instead there is a large spread between the 10th and 90th percentiles of a given teacher characteristic, this reflects a nonuniform distribution of teachers and may provide evidence of teacher sorting or systematic assignment of teachers to certain types of schools.
Research Question 2: Does the Distribution of Teacher Qualifications in each Country Favor less or more Advantaged Children?
The next step in the sorting analysis is to determine whether teacher sorting occurs according to school and community characteristics. If the distribution of teachers is uniform, we should not see large differences in teacher characteristics across urban and rural schools, for example. In contrast, large differences in teacher characteristics across school types suggest teacher sorting. To explore whether the distribution of teachers favors more or less advantaged students, we describe average teacher characteristics across community size and average parental education. We also correlate school-average teacher characteristics with 3 measures of school working conditions: material shortage, student delinquency, and teachers working morale. Strong relationships between teacher characteristics and school working conditions can also provide evidence of teacher sorting.
After conducting these analyses in both Mexico and South Korea, we explore differences in patterns across the two countries. If differences in teacher characteristics across school types are more pronounced in one of the two countriesand if these differences reflect greater access to more qualified teachers for more advantaged childrenthis provides evidence of a less equitable distribution of teachers. Again, given the relatively centralized teacher assignment system in both Mexico and South Korea, evidence that one system is more equitable than another suggests the influence of other factors, which could include a greater culture of equity or differences in teacher-related policies, such as periodic rotation of teachers or incentives to teach disadvantaged populations.
Research Question 3: Assuming Differences in the Distribution of Teacher Qualifications across Mexico and South Korea, How Can Dissimilarities in Teacher-Related Policies and Educational Priorities Help to Explain these Differences?
Our analysis of the distribution of teachers leads to a discussion of factors that may be related to differences between Mexico and South Korea. We draw from a number of OECD reports that provide recent and reliable measures of educational contexts in Mexico and South Korea, including the OECDs 2005 Teachers Matter report, its 2010 Education at a Glance report, and results of the 2009 PISA. We explore differences in social and educational contexts, the distribution of school resources, and teacher-related policies in the two countries. Although we cannot identify causal relationships between these factors and the distribution of teachers in both countries, this analysis can shed light on key factors to be considered by policy makers and researchers.
Tables 1 and 2 provide descriptive statistics of the TALIS variables that are relevant to our study. To begin with, there are small differences between Mexico and South Korea in school autonomy related to teacher hiring (Table 1). This autonomy can be measured in two ways in TALIS: first, based on the percentage of principals of lower secondary schools who reported that schools have considerable autonomy in selecting teachers for hire. According to this measure, South Korean principals have somewhat more autonomy (31%) than principals in Mexico (24%). Second, according to an index of school autonomy in hiring and firing teachers and setting teacher salaries, as reported by the school principal, Mexico and South Korea have similar scores, -0.68 and -0.64, respectively.4
The South Korean sample is more urbanized, with a third of respondents in large cities, compared to 22% in Mexico (Table 2). Mexico has a higher share of respondents in private schools (26% compared to 21% in South Korea). In terms of school conditions, South Korea has higher levels of parental education, especially the percentage of parents with more than or equal to upper secondary education (ISCED Level 3). Mexico has substantially higher levels of material shortage and student delinquency than South Korea, but teachers also report having a higher level of working morale.
Mexican teachers are more likely to be male (47% to 36%), less likely to have more than a bachelors degree (11% compared to 35%), and more likely to have less than a bachelors degree (13%) than teachers in South Korea (only 1%). Teacher experience levels across the two countries are similar, as is the percentage of teachers who have received training in their subject. Finally, Mexican teachers score substantially higher in terms of self-efficacy beliefs relative to South Korean teachers. In fact, South Korea ranks at the bottom of all TALIS countries in the self-efficacy measure.5 While this is consistent with prior research that non-Western cultural groups tend to have lower efficacy beliefs, research has also pointed out that self-efficacy beliefs can be confounded with cultural notions of individualism versus collectivism (e.g., Klassen, 2004). Additionally, according to the OECDs (2009) TALIS report, because there were differences between countries in the structure of the efficacy index, small differences between countries may be due to factors specific to the countries themselves. As a result, we must take care in comparing self-efficacy measures across countries.
THE DISTRIBUTION OF TEACHER QUALIFICATIONS IN MEXICO AND SOUTH KOREA
Our first research question is whether in each country, teacher qualifications are distributed uniformly across schools. Table 3 provides school-level averages of teacher characteristics at the 10th, 50th, and 90th percentiles in both Mexico and South Korea. In both countries, teacher characteristics are not uniform across schools, illustrated by substantial differences between the 10th and 90th percentiles in education, experience, and self-efficacy.6 In Mexico, the 10th percentile score for teachers with more than 20 years of experience is zero, indicating that in at least 10% of schools, no teacher has this level of experience. In contrast, at the 90th percentile of this variable, 63% or more of teachers have 20 years or more of experience. A smaller but notable spread occurs in the percentage of Mexican teachers with greater than a bachelors degree: at the 10th percentile, no teacher has this level of educational attainment, while at the 90th percentile, nearly a quarter of teachers have a bachelors degree or higher.
In South Korea, there is also evidence of an uneven distribution of teachers, with sizeable differences between the 10th and 90th percentiles in teacher education and experience. Most notably, schools at the 10th percentile in the percentage of highly educated teachers have only 13% of teachers with greater than a bachelors degree, compared to nearly two thirds of teachers at the 90th percentile. Comparing the two countries, Mexico has larger differences between the 10th and 90th percentiles in the percentage of teachers with less than a bachelors degree, the percentage with more than 20 years of experience, and self-efficacy. South Korea has a larger spread between the 10th and 90th percentiles in the percentage of male teachers, the percentage with greater than a bachelors degree, and the percentage of novice teachers. Given these results, it is difficult to determine which country has a less uniform distribution of teachers. This information also does not indicate whether one system is more equitable than the other, because it does not tell us whether teacher characteristics vary across specific community or school types. We explore this question in the analysis that follows.
THE DISTRIBUTION OF TEACHER QUALIFICATIONS ACROSS COMMUNITIES AND SCHOOLS
To explore our second research question, we examine school-level averages of teacher characteristics according to community and school characteristics. Specifically, we examine teachers gender, educational attainment, experience, training in their subject, and self-efficacy across community size and average parental education in a school. We also explore correlations between school-level teacher characteristics and school working conditions.
In this analysis, we examine differences in teacher qualifications between large cities and the other community size groups (village, small town, town, and city).7 Table 4 indicates that Mexican teachers in large cities have higher average education levels, in terms of the percentage with more than a bachelors degree (12%), relative to teachers in villages (10%) and small towns (7%). However, the difference between large cities and villages is not statistically significant. The percentage of teachers with greater than 20 years of experience is 38% in large cities, compared to 26% in villages, a statistically significant difference. Teachers self-efficacy levels are higher in large cities than in towns, but across the 5 community size categories, teachers in Mexican villages report the highest levels of self-efficacy. However, the difference in self-efficacy between villages and large cities is not statistically significant.
Patterns across community size in South Korea indicate an advantage for students in villages and small towns, relative to large cities. Teachers in villages are substantially and significantly more likely to have greater than a bachelors degree compared to teachers in large cities (77% to 32%). Teachers in both villages and towns are also more likely to have more than 20 years of experience (45% and 44% respectively) than teachers in large cities (30%), but the difference between villages and large cities is not statistically significant. The percentage of teachers with training in their subject is significantly higher in villages than in large cities. Finally, teachers in villages report higher levels of self-efficacy than teachers in large cities, but this difference is not statistically significant.
Comparing Mexico and South Korea in terms of qualifications of teachers across schools, the pattern in Mexico is mixed with a small advantage for urban areas. Relative to villages, large cities have a small but statistically significant advantage in terms of teacher education, and a larger and significant advantage in the percentage of veteran teachers. In contrast, South Korean students living in smaller communities have a clear advantage in terms of teacher education level and experience. Figures 1 and 2 provide visual representations of these cross-national differences in the distributions of teacher education and experience respectively. Figure 1 also demonstrates the large teacher education advantage held by South Korea, with respect to the overall percentage of teachers with greater than a bachelors degree. In Mexico, we see an upward trend in both education and veteran teachers as community size increases, whereas South Korea illustrates the reverse trend. Assuming that students in villages and small towns are on average less advantaged than students in large cities (discussed below), these results support the negative (equity-enhancing) opportunity gap that Akiba et al. (2007) found in South Korea.
Table 5 reports teacher qualification percentages according to the school-level percentages of parents completing upper secondary education (ISCED 3) or higher. Assuming that the children of more educated parents are higher-achieving, and that teachers prefer to work with higher-achieving students, then we would expect to see more qualified teachers concentrated in schools with higher average parental education, all else equal. In Mexico, we see few statistically significant differences between schools with the most educated parents (60% or higher have greater than an upper secondary degree) and the least educated parents (10% or fewer have an upper secondary degree). Two exceptions are in the percentage of male teachers, which is significantly higher in schools with the lowest parental education, and the percentage of teachers with 20 or more years of experience, which is significantly higher in schools with the highest parental education (49%) compared to schools with the lowest parental education (34%). In South Korea, we see a clear advantage in teacher qualifications for the lowest parental education category relative to the highest: the percentage of teachers with bachelors degrees is 20 percentage points higher among the lowest group. The percentage of new teachers is 0% in the lowest group, compared to 7% of teachers in the highest parental education group. Both of these differences are statistically significant. Additionally, the percentage of teachers with 20 years or more of experience is 12 percentage points higher in high-parental education schools relative to low-parental education schools, although this difference is not statistically significant.
Overall parental education levels are substantially higher in South Korea relative to Mexico, so the low-education category in South Korea is much smaller and may be less representative of South Korean schools; as indicated in Table 2, only 5% of TALIS schools in South Korea have fewer than 10% of parents who have completed upper secondary school. The corresponding percentage in Mexico is 26%. Nonetheless, Table 5 reveals a general trend in South Korea of increasing teacher educational attainment and experience as parental education decreases, as illustrated in Figures 3 and 4. In other words, the distribution of teacher qualifications in South Korea appears to favor children with less educated parents. Figures 3 and 4 illustrate the reverse trend in Mexico.
Correlations between Teacher Characteristics and School Working Conditions
To shed further light on the distribution of teachers, we correlated school-average teacher characteristics with measures of school working conditions (Table 6). The working condition variables include indices of material shortage, student delinquency, and teachers working morale.8 These correlations can provide information regarding whether less or more qualified teachers tend to work in more difficult environments. For example, a positive correlation between the percentage of novice teachers in a school and the level of material shortage could indicate that new teachers are disproportionately concentrated in schools with fewer resources. In Mexico there are few significant correlations: the percentage of male teachers is positively correlated with material shortage, while the percentage of teachers with more than 20 years of experience is negatively correlated with material shortage and positively correlated with working morale. In South Korea, we see more significant correlations. The percentage of novice teachers in a school (0-2 years of experience) is positively correlated with both student delinquency and working morale, while the percentage of teachers with more than 20 years is negatively correlated with working morale. The percentage of teachers with training in their subject is positively correlated with material shortage. Average teacher self-efficacy is negatively correlated with both delinquency and working morale.
This analysis does not allow us to determine whether correlations indicate sorting, or an actual influence of a certain type of teacher on a school. For example, are novice South Korean teachers more likely to be assigned to schools with high levels of delinquency, or does the presence of many inexperienced teachers result in more delinquent behavior? However, it is difficult to argue that teachers can cause material shortages. Examining correlations of this variable with teacher characteristics, there is evidence that veteran teachers in Mexico are at an advantage, while South Korean teachers with training in their subject are at a disadvantage. This provides further, albeit limited, evidence of an equity-enhancing distribution of teachers in South Korea, compared to an inequitable distribution of teachers in Mexico.
EXPLORING CROSS-NATIONAL DIFFERENCES IN THE DISTRIBUTION OF TEACHER QUALIFICATIONS
According to our analysis, the distribution of teachers is uneven in both Mexico and South Korea. In the Mexican case, the distribution of teacher experience and education tends to favor larger communities and schools with better-educated parents. In South Korea, the reverse is true. What explains these differences? Our conceptual framework posits that the level of centralization and commitment to equity in an educational system are key factors in determining the allocation of teachers across schools. Because Mexico and South Korea are similar in terms of the level at which teachers are hired and assigned, this discussion focuses primarily on issues related to the nations commitment to equity, which is reflected in social and educational contexts, the allocation of educational resources, and teacher-related policies. We discuss these key factors and relationships below. We also discuss how differences in contexts and policies may be related to cross-national differences in national educational performance.
Social and Educational Context
To the extent that teachers can select where they teach based on their preferences for student composition, the underlying distribution of the student population should be related to the distribution of teachers. Specifically, greater similarities in student populations across schools will lead to smaller cross-school differences in teacher quality. Although teachers ability to select assignments based on student composition is limited in Mexico and South Korea, in more decentralized systems (like the United States), the underlying distribution of students is likely to have a strong influence on the distribution of teachers. Additionally, heterogeneity across schools in student socioeconomic status and educational performance provides an important measure of a nations commitment to equitable opportunities and resources for all students, regardless of background.
Examination of social and educational indicators suggests major differences in the distribution of educational resources between Mexico and South Korea. For example, the Gini index, which describes the extent to which citizens share in national income, is much lower in South Korea. In 2009, South Koreas Gini was 31.4, compared to 48.2 in Mexico in 2008 (CIA, 2010), indicating that South Korea has substantially greater income equality. In terms of educational equity, the results of the 2009 PISA reveal many differences between Mexico and South Korea in student composition and resources. To begin with, 15-year-old Mexican students have a much wider range of socioeconomic status between the 5th and 95th percentiles than in South Korea, indicating a greater gap in SES between advantaged and disadvantaged students and a more socioeconomically heterogeneous student population (OECD, 2010c, Figure 2.3.2). This is despite the fact that a much greater percentage of South Korean 15-year-olds remain in the school system, compared to Mexico. In other words, although about 40% of Mexican 15-year-olds are not in school, those who are in school are much more socioeconomically heterogeneous than students in South Korea and all other members of the OECD (OECD, 2010c, Table 2.3.2).
The Allocation of School Resources
The allocation of teachersthe most important of educational resourcesdoes not occur in isolation from the allocation of other school resources. In other words, if the distribution of qualified teachers is skewed toward more advantaged students, we are likely to see a similar distribution of other school resources. Results of the 2009 PISA indicate some differences between South Korea and Mexico in school-level correlations between student SES and school resources. In most cases, these correlations illustrate more equitable allocations in South Korea. In Mexico, correlations indicate that more advantaged students have more computers per student and higher-quality school resources. In both cases, correlations are significantly different from the OECD average. In contrast, disadvantaged students in South Korea have an advantage in terms of the number of computers per student and lower pupil-to-teacher ratios. The correlation between student SES and quality of school resources is not statistically significant in South Korea (OECD, 2010c, Figure 2.2.3).
As we indicate above, teacher hiring decisions are made relatively centrally in both Mexico and South Korea. According to the TALIS data, 31.2% of South Korean principals reported considerable autonomy at the school level for hiring teachers, compared to 24.3% of Mexican principals (Table 1). Autonomy measures for both countries fall significantly below the TALIS average of 67.7%, indicating that school leaders in both countries exercise relatively little autonomy in teacher hiring compared to the rest of the TALIS sample. According to the OECDs (2005) Teachers Matter report, teacher recruitment and hiring in South Korea occur relatively centrally. In contrast to countries like the United States, where school recruitment and hiring take place at the local education authority or the school, teachers in South Korea are recruited and hired at the regional school authority level (OECD, 2005). South Korea uses a competitive examination whereby teacher candidates take a national exam; those who score above a specific benchmark are selected for the teacher candidate pool. These candidates then undergo a second stage evaluation consisting of a written test, an interview, and an assessment of teaching ability. Successful candidates are then assigned to schools by metropolitan or provincial offices of education (Coolahan, Santiago, Phair, & Ninomaya, 2004). Every five years, teachers are required to rotate to a new school within their city or province (Kang & Hong, 2008).
In Mexico, teacher hiring and placement occur at the state level, rather than at the school or district (Guevara & González, 2004). Teachers can choose the state where they teach, but not the city or school. Teacher assignment practices vary considerably across Mexicos 32 states. Thirteen states apply a selection mechanism to identify and hire teachers while 19 states have no selection mechanism for prospective teachers. Once teachers have been hired, the process by which they are placed in schools also varies from state to state and between newly created positions and those that are vacated due to the retirement or transfer of another teacher. To fill newly created positions, state education officials have sole decision-making authority in 11 states and the teachers union assigns teachers to schools in 3 states. In the remaining 18 states, the state education authority and union each fills half of the available positions. In filling vacated posts, unions tend to have greater authority: state representatives of the union make assignment decisions in 13 states, the state education agency makes the decision in 7 states, and teacher assignment is evenly decided in 12 states (Guevara & González, 2004).9
In Mexico, the teacher hiring and assignment process varies considerably across states, but in South Korea it is much more standardized. In both countries, teacher-hiring decisions are largely out of the hands of school leaders and local education agencies. More central actors choose which teachers to hire and where to place them. As we argue above, to some extent, this allows us to control for level of centralization across the two countries and to examine whether other factorssuch as a culture of equitymay influence the distribution of teachers across countries.
A commitment to equity can be inferred in part from other practices and policies related to teachers assignments. For example, the use of incentives to encourage teachers to work with disadvantaged students indicates policy makers concern to ensure an equitable distribution of teachers. Both Mexico and South Korea have such incentives in place. While Mexican teachers work under a uniform salary schedule, those who teach in very poor areas can receive extra compensation. Additionally, rural teachers who attend their schools regularly receive some financial support from a federal initiative to improved educational conditions in low-achieving areas (Guevara & González, 2004). Teachers who work in low development zones and participate in Mexicos national teacher incentive program, Carrera Magisterial, can also advance through the system more quickly. South Korea also provides incentives for teaching in areas with disadvantaged student populations. These include smaller classes, less classroom teaching time, stipends, and advantages in choosing future positions and in seeking promotion (Kang & Hong, 2008).
Because incentives in both Mexico and South Korea are offered in the form of both salary bonuses and promotional opportunities, it is difficult to quantify and compare the relative size of the incentives offered. Although a comparison of the relative size of teacher incentives across the two countriesboth in terms of salary and promotionwould be very valuable, such an analysis is beyond the scope of this study. Moreover, there is little evidence regarding whether and how well these incentives work to attract and retain qualified teachers in schools with disadvantaged children. One way to examine this question is by examining teacher absenteeism rates across community sizes in the two countries. To the extent that incentive programs serve to attract and retain teachers, we would expect to see lower absenteeism rates in areas where incentives are offered. According to TALIS data, the percentage of schools reporting that teacher absenteeism hinders student learning to some extent or a lot varies substantially across community size in both Mexico and South Korea. In South Korea, no schools in villages report problems with teacher absenteeism, compared to 23% of schools in small towns, 17% in towns, 19% in cities, and 27% in large cities. In Mexico, teacher absenteeism appears to be much more severe, but as in South Korea, schools in Mexican villages report the lowest severity of teacher absenteeism (44%). The percentage of Mexican schools reporting that teacher absenteeism hinders student learning is higher in small towns (66%), towns (69%), cities (52%), and large cities (75%). This limited evidence may indicate that incentives help to reduce absenteeism (and perhaps retain teachers) in villages.10
An important distinction between South Korea and Mexico is the use of teacher rotation in South Korea. As discussed above, South Korean teachers must move to a new school within their city or province every five years, in order to prevent the concentration of higher-quality teachers in certain schools. Kang and Hong (2008) argue that this rotation of teachers, coupled with incentives to teach disadvantaged children and uniformly high-quality teachers, helps to explain why the distribution of teachers in South Korea favors disadvantaged children. Yet Mexican teachers also receive incentives to teach disadvantaged children. Additionally, uniformity in teacher quality should result in a uniform distribution of teacher quality, rather than an equity-enhancing distribution. The key policy difference between South Korea and Mexico appears to be the periodic rotation of teachers across schools in South Korea. Although we cannot draw a causal link between this policy difference and the distribution of teachers in each country, the rotation of teachers may help to explain some of the results we find in our analysis of the TALIS data. In particular, rotation of teachers in South Korea appears to prevent the concentration of very experienced teachers in larger communities and in schools with high levels of parental education, as we see in Mexico.
Ample evidence indicates that one of the things that teachers consider when choosing their schools is the performance of students (e.g., Lankford et al., 2002). As such, student performance can be considered to be a determinant of teacher sorting patterns. At the same time, the distribution of teachers can also influence student performance, which is ultimately why it concerns educational researchers and policy makers. Disproportionate concentrations of highly qualified teachers among advantaged children are likely to result in pronounced achievement gaps between low- and high-SES students (Darling-Hammond, 2004). Students access to qualified teachers may also influence the degree to which their social class influences their achievement, and whether students are able to succeed despite being born in poverty or without significant educational resources in the home.
To explore how differences in the distribution of teachers may influence academic performance in Mexico and South Korea, we briefly examine the performance of 15-year-olds in these two countries on the 2009 PISA. In reading performance, the slope of the socioeconomic gradient, or the relationship between students socioeconomic status and their test performance, is actually lower in Mexico (25) than in South Korea (32), but both countries fall below the OECD average of 38, indicating greater than average educational equity in both countries (OECD, 2010c, Figure 2.1.4).11 However, variations in students socioeconomic background explain a greater percentage of the variation in reading achievement in Mexico (14.5%) relative to 11% in South Korea (OECD, 2010c), indicating that the relationship between student SES and test performance is stronger in Mexico than in South Korea. While the marginal change in performance associated with changes in SES is larger in South Korea, the relationship between SES and performance is stronger in Mexico. This suggests that in South Korea, other factors not related to student SES play a more central role in improving student learning.
Although socioeconomic gradients provide mixed results regarding the relative equity in each system, according to PISA results, disadvantaged South Korean students are much more resilient than disadvantaged Mexican students, meaning that they are more likely to perform better than predicted given their low socioeconomic status. In 2009, nearly 60% of disadvantaged students performed better than predicted in South Korea, compared to less than 30% in Mexico and the OECD average of 35% (OECD, 2010c, pp. 62-63).12 Moreover, the OECD (2010c) identifies South Korea (along with Canada, Finland, and Hong Kong) as places that enjoy both high levels of performance and high levels of equity. In contrast, Mexico has lower-than-average performance, coupled with gentler-than-average (p. 60) gradient slopes. The OECDs 2009 PISA report also points out that because PISA tests only 15-year-olds who are in school, the impact of socioeconomic background on the reading performance of 15-year-olds may be underestimated where enrollment rates are low (OECD, 2010c, p. 60). As Mexico enrolls only about 60% of its 15-year-old students in school, there is clearly a need to take caution in interpreting these results.
Perhaps the most striking differences in reading performance between Mexico and South Korea are urban/rural gaps that are mirror images of each other. After controlling for student socioeconomic status, Mexican students in urban areas outscore students in rural areas by 50 points, or half of a standard deviation, the fourth highest urban-rural disparity among OECD countries, after Hungary, the Slovak Republic, and Turkey. In contrast, South Korean students in rural schools outperform urban students by 44 points after adjusting for student socioeconomic status, making it one of the few countries in PISA with a large rural performance advantage (OECD, 2010c). Despite this advantage, several sources indicate that students in rural areas of South Korea have fewer resources than students in urban areas, including lower family economic status and less access to tutors and private institutions (Choi et al., 2003; Hang & Kong, 2008).
Although we cannot draw a causal relationship, it seems very likely that South Koreas rural advantage is linked to the large differences in teacher education and experience in South Korea that we see in Table 4 and Figures 1 and 2. In particular, South Korean villages appear to have a distinct advantage in terms of the qualifications of their teachers. Additionally, teacher absenteeism, as reported in TALIS, appears to be much less of a problem in smaller communities relative to larger communities in South Korea, which could also influence differential performance of South Korean students in PISA. In contrast, the dismal performance of rural Mexican students on PISA may be related to opposite patterns of teacher qualifications, although the lack of statistically significant differences in teacher characteristics between large cities and villages makes this a tentative conclusion at best. Moreover, although teacher absenteeism appears to be less of a problem in Mexican villages than in larger communities, absenteeism is nonetheless quite high in villages, with 44% of school principals in Mexican villages reporting that teacher absenteeism hinders student learning to some extent or a lot, compared to none in South Korea.
In sum, South Korea has a more homogeneous student population, disadvantaged students who are more likely to achieve better than expected, and an achievement gap that favors rural students, after controlling for student SES. In cases where correlations between student SES and school resources are significant, these correlations provide evidence of advantages for lower-income children in South Korea. In Mexico, both resource quality and computers per pupil favor more advantaged children. Taken as a whole, differences between Mexico and South Korea help to explain the equity-enhancing patterns of qualified teachers in South Korea and inequitable patterns in Mexico. We argue that these differences also influence performance differences across the two countries.
DISCUSSION AND CONCLUSION
Our primary research question was whether less advantaged children in Mexico and South Korea have uniform access to well educated, experienced, and efficacious teachers. In both countries, the distribution of teachers across schools is not uniform, but the equity implications of an uneven distribution are very different. In Mexico, the distribution of teacher qualifications across schools places lower-SES students at a disadvantage. In South Korea, we see substantial evidence that more disadvantaged students have greater access to better educated and more experienced teachers. Thus, evidence from TALIS provides moderate support for our hypothesis that, relative to South Korea, the distribution of teacher qualifications is less equitable in Mexico. Our results from South Korea also confirm the presence of a negative opportunity gap in terms of disadvantaged childrens access to qualified teachers (Akiba et al., 2007).
What explains the equity-enhancing distribution of teachers in South Korea? It is likely that a number of policy and institutional influences come into play. First, a relatively centralized system ensures that education officials can place teachers according to need. At the same time, this centralization reduces teachers ability to make decisions based on their own preferences for working conditions. In other words, supply-side factors become less important and demand-side factors increase in prominence. Second, a number of explicit policies, such as the mandatory periodic rotation of teachers across schools and incentives to teach in schools with disadvantaged students, also are likely to play a role. Third, a general commitment to equity in society and education, as evidenced by the social and economic indicators discussed above, ensure that these policies and practices are not only developed, but also implemented and enforced.
A final factor that may help to explain cross-national differences is that the underlying distribution of studentsas measured by the range of socioeconomic status between the 5th and 95th percentiles of 15-year-old students participating in PISAis much more homogeneous in South Korea than in Mexico, so to the extent that teachers respond to student characteristics, there is less diversity in what they respond to. One may argue that given this important difference in student populations, other policies and practices may not have a measurable impact on the distribution of teacher qualifications. On the other hand, large heterogeneity in the Mexican student population makes explicit policies to address the teacher qualification gap even more critical.
Given much lower levels of social and educational equity in Mexico, differences in teacher characteristics across community size and average parental education could be much starker than what we find in the TALIS data. It is possible that specific policies and arrangements, such as a relatively centralized teacher hiring and assignment system and incentives to teach in remote and rural areas, attenuate tendencies toward teacher sorting. It is also possible that patronage and bargaining among teachers leads to teacher assignments that are based on political connections with influential education or union officials (Martin, 1993). If this is the case, then significant teacher sorting could occur, but we would not see it reflected in the data, unless political connections are correlated with teacher qualifications. Finally, the TALIS data may simply not provide us with the range of teacher qualification measures that we need to identify clear teacher sorting patterns. Other studies of teacher sorting in Mexico that use richer teacher quality measures including experience, education levels, and teacher test scores and training scores, have found significant teacher sorting effects both within and across Mexican states (Luschei, 2012).
These results have important implications for understanding cross-national differences in both the distribution of teachers and student performance. Ultimately, the distribution of teachers is important not just because it reflects national priorities, but also because it probably influences student performance and SES-based achievement gaps. Large performance differences between Mexico and South Korea in PISA, especially differences in the performance of urban and rural students, provide evidence of the importance of allocating teachers equitably.
LESSONS FOR THE UNITED STATES
Most of the empirical research related to the distribution of teachers has been conducted in the United States. This research clearly demonstrates an inequitable distribution of teachers that places poor, minority, and low-achieving students at a disadvantage. ur framework suggests that the inequitable distribution of U.S. teachers results at least in part from a high level of social inequality (as evidenced by a Gini index of 45 in 2007), coupled with considerable school-level autonomy in hiring teachers. Unfortunately, the United States did not participate in the initial round of TALIS, so we cannot compare the distribution of U.S. teachers to teachers in TALIS countries. Among countries participating in TALIS, Portugal is most like the United States in terms of having both high autonomy (81.3% of principals report having considerable responsibility for selecting teachers for hire) and relatively high inequality (a Gini index of 38.5 in 2007). Analyses of TALIS data from Portugal (not reported here but available upon request) produce results similar to those of teacher sorting research in the United States. These results demonstrate a clear pattern in which more experienced teachers (more than 20 years of experience) are more likely to teach in larger communities and in schools with higher average parental education. Better educated teachers are also more likely to teach in large cities, relative to small towns. An unequal distribution of experienced and well educated teachers could reflect greater bargaining power of these teachers, who, due to high levels of teacher autonomy, are able to negotiate directly with schools. As in the United States, there is also a relatively high degree of inequality in Portugal, which could result in the lack of policies to stem teacher sorting behavior.
LIMITATIONS AND AREAS FOR FUTURE RESEARCH
Because we examine two countries with relatively centralized teacher assignment systems, we cannot test our hypothesis that centralization leads to greater equity in students access to qualified teachers. Other researchers have argued for greater centralization of educational decision making to provide more equitable opportunities for students. For example, Carnoy and Marshall (2005) argue that Cubas superior academic performance among Latin American nations is due in part to a high degree of centralization and the states ability to generate social capital for families and children. An analysis involving a greater diversity of countries, with varying degrees of centralization, would help to shed light on this question. In the event that future research does confirm our hypothesis, policy makers will need to make difficult decisions about the benefits of educational centralization versus school and teacher autonomy. Although we hypothesize that greater autonomy leads to more inequity, we also recognize the many advantages that autonomy provides to schools and teachers. In an analysis of TIMSS data, Woessmann (2003) found that students in schools with greater autonomy in making decisions about teacher hiring performed better in both mathematics and science. Additionally, the OECD (2005) suggests that greater autonomy in hiring teachers can result in a better match between schools and their teachers. At the same time, we must ask ourselves if that match is most likely to occur in schools that can easily attract qualified teachers due to their superior working conditions.
Finally, despite the richness of the TALIS data, we are only able to examine a limited number of teacher and school characteristics. As we note above, other studies of teacher sorting in Mexico have found stronger evidence of teacher sorting both within and across Mexican states (Luschei, 2012). This raises the question of whether the same could be said for South Korea, and whether we can use our analysis of the TALIS data to make recommendations for changes in policy and practice. We know of no other study of the distribution of teachers in South Korea that we can compare our results with. It is quite possible that real sorting in South Korea is much greater than we have found, and we cannot claim to have definitively characterized the degree of teacher sorting in either country. We encourage additional comparative exploration of these questions both within and across countries with diverse data sources and measures of teacher qualifications. We believe that the central question guiding such analysis should be how policy makers, educators, and society can ensure equitable (or equity-enhancing) access to qualified teachers for all children, regardless of their personal and family backgrounds.
1. Akiba et al. (2007) found that in South Korea, the overall teacher quality opportunity gap is -3.9%, indicating that the percentage of low-SES students taught by qualified teachers is nearly 4% greater than the percentage of high-SES students taught by qualified teachers. The United States gap is 14.4%, indicating a distribution of teachers that favors high-SES children. The United States has the fourth highest gap among 38 participating countries, behind Syria (18.5%), Chile (17.1%), and Taiwan (16.8%). The opportunity gap in Japan is 0.9%.
2. One exception is Hill (2007), who assessed teachers mathematical knowledge for teaching, and then examined the relationship between teachers mathematics expertise and student socioeconomic status. She found that higher-knowledge teachers are more likely to teach affluent students.
3. Hanson (2006) notes that if private contributions to education are included, the percentage of South Koreas national income spent on education increases to 8.2%, the highest among OECD countries and considerably higher than Mexicos 5.9%. According to Hanson, South Koreas sizeable private investment in education signals the enormous sacrifice South Korean families are willing to make to assure the education of their children (p. 641).
4. The school autonomy index was derived by the TALIS research team using principal components analysis of 5 items: selecting teachers for hire, firing teachers, establishing teachers starting salaries, determining teachers salary increases, and allocating funds for teachers professional development. A higher value indicates higher levels of school responsibility. Among participating countries, the school autonomy index ranges from -1.03 in Malaysia to 1.42 in Norway. The TALIS average is -0.04 (OECD, 2010d, 2010e).
5. The self-efficacy index was derived by the TALIS research team using multi-group confirmatory factor analysis of 4 items from the teacher questionnaire related to whether they felt they made a difference in students lives, whether they could make progress with difficult and unmotivated students, how successful they were with their students, and whether they felt they knew how to get through to their students. Among participating countries, the self-efficacy index ranges from -0.77 in South Korea to 0.51 in Norway. The TALIS international average is 0 (OECD, 2009).
6. The nature and distribution of the training variable do not allow for the calculation of a value at the 90th percentile.
7. TALIS designates community types according to the size of the population in each. Villages (also known as hamlets or rural areas) have fewer than 3,000 people; small towns have 3,000 to about 15,000 people; towns have 15,000 to about 100,000 people; cities have 100,000 to about 1,000,000 people; and large cities have over 1,000,000 people (OECD, 2010e).
8. The material shortage index was derived by TALIS using principal components analysis. It consists of 4 items pertaining to the shortage or inadequacy of instructional materials computers for instruction, other equipment, and library materials. The material shortage index has acceptable reliability for South Korea (α = 0.943) and Mexico (α = 0.885) (OECD, 2010d). The student delinquency and teacher working morale indices were derived by TALIS using multi-group confirmatory factor analysis. Student delinquency consists of 6 items (vandalism, theft, intimidation or verbal abuse of other students, physical injury to other students, intimidation or verbal abuse of teachers or staff, and use/possession of drugs and/or alcohol). Teacher working morale consists of 3 items (arriving late at school, absenteeism, lack of pedagogical preparation). They have acceptable reliability for both South Korea (α = 0.947 and α = 0.909) and Mexico (α = 0.955 and α = 0.949) as well as model fit and cross-cultural validity (although the comparison of means across countries requires careful interpretation) (OECD, 2010d).
9. In 2008, an agreement between the Mexican Secretariat of Public Education and the teachers union, known as the Alliance for Educational Quality, established a new system, whereby teacher candidates take a national exam and are placed in schools according to their results (SEP, 2011). This reform may serve to make teacher assignment, at least for new teachers, more uniform across the Mexican states.
10. The teacher absenteeism variable provides an interesting direction for future research on teacher sorting. In our sorting analysis we did not include the absenteeism variable because it represents a teacher behavior rather than a teacher characteristic or qualification, which are the primary foci of our analysis.
11. The gradient can be interpreted as the increase in test score points associated with a one standard deviation increase in PISAs index of economic, social, and cultural status. For example, the OECD average indicates that across OECD countries, test scores increase by 38 points for each one standard deviation increase in the index (OECD, 2010c, p. 54). According to the OECD (2010c), the slope of the gradient is shown by the inclination of the gradient line: the sharper the inclination, or the closer it is to a vertical line, the greater the impact of economic, social and cultural status on student performance, suggesting greater inequity; gentler gradients indicate a lower impact of socio-economic background on student performance, i.e. more equity (p. 54).
12. The socioeconomic gradient describes the overall relationship between students socioeconomic status and their test performance. However, in every country there are students whose performance varies from the gradient. Disadvantaged students who perform better than predicted by the socioeconomic gradient are considered to be resilient.
Aaronson, D., Barrow, L., & Sander, W. (2007). Teachers and student achievement in the
Chicago public high schools. Journal of Labor Economics, 25(1), 95-135.
Akiba, M., LeTendre, G. K., & Scribner, J. P. (2007). Teacher quality, opportunity gap, and national achievement in 46 countries. Educational Researcher, 36(7), 369-387.
Bacolod, M. (2007). Who teachers and where they teach: College graduates of the 1990s. Educational Evaluation and Policy Analysis, 29(3), 155-168.
Boyd, D., Lankford, H., Loeb, S., & Wyckoff, J. (2003). Understanding teacher labor markets: Implications for educational equity. In M. Plecki & D. Monk (Eds.), School Finance and Teacher Quality: Exploring the Connections. Larchmont, NY: Yearbook of the American Education Finance Association, Eye on Education.
Boyd, D., Lankford, H., Loeb, S., & Wyckoff, J. (2005). The draw of home: How teachers preferences for proximity disadvantage urban schools. Journal of Policy Analysis and Management 24(1), 113-132.
Central Intelligence Agency [CIA]. (2011). CIA world factbook. Retrieved from https://www.cia.gov/library/publications/the-world-factbook/rankorder/2172rank.html?countryName=Mexico&countryCode=mx®ionCode=na&rank=28#mx
Carnoy, M., & Marshall, J. (2005). Cubas academic performance in comparative perspective. Comparative Education Review, 49(2), 230-261.
Choi, S., Kim, Y., Ryu, H., Kim, H., Lee, H., & Lee, J. (2003). Analysis of private tutoring status and its expense scale. Seoul: Korean Educational Development Institute.
Clotfelter, C. T., Ladd, H. F., & Vigdor, J. L. (2005). Who teaches whom? Race and the
Clotfelter, C. T, Ladd, H. F., & Vigdor, J. L. (2006). Teacher-student matching and the assessment of teacher effectiveness. Journal of Human Resources, 41(4), 778-820.
Coolahan, J., Santiago, P., Phair, R., & Ninomiya, A. (2004). Attracting, developing and retaining effective teachers. Country note: Korea. Paris: OECD.
Darling-Hammond, L. (2004). Inequality and the right to learn: Access to qualified teachers in Californias public schools. Teachers College Record, 106(10), 19361966.
Ferguson, R. F. (1991). Paying for public education: New evidence on how and why
money matters. Harvard Journal on Legislation, 28(2), 458498.
Ferguson, R. F., & Ladd, H. F. (1996). How and why money matters: An analysis of Alabama schools. In H. F. Ladd (Ed.), Holding schools accountable: Performance-based reform in education (pp. 265-298). Washington, DC: The Brookings Institution.
Goldhaber, D. (2002). The mystery of good teaching: Surveying the evidence on student achievement and teachers characteristics. Education Next, 2(1), 50-55.
Goldhaber, D. D. & Brewer, D. J. (2000). Does certification matter? High school teacher
certification status and student achievement. Education Evaluation and Policy Analysis, 22(2), 129-145.
Greenwald, R., Hedges, L. V., & Laine, R. D. (1996). The effect of school resources on student achievement. Review of Educational Research, 66(3), 361−396.
Guevara, M. R., & González, L. E. (2004). Attracting, developing and retaining effective teachers: Country background report for Mexico. Paris, France: OECD.
Hanson, M. (2006). Transnational corporations as educational institutions for national development. The contrasting cases of Mexico and South Korea. Comparative Education Review, 50(4), 625-650.
Hanushek, E. A. (1986). The economics of schooling: Production and efficiency in public schools. Journal of Economic Literature, 24(3), 11411177.
Hanushek, E. A. (1997). Assessing the effects of school resources on student performance: An update. Educational Evaluation and Policy Analysis, 19(2), 141-164.
Hanushek, E. A., & Rivkin, S. G. (2004). How to improve the supply of high-quality teachers. In D. Ravitch (Ed.), Brookings papers on education policy 2004 (pp. 7-44). Washington, DC: Brookings Institution Press.
Hill, H. C. (2007). Mathematical knowledge of middle school teachers: Implications for the No Child Left Behind Policy Initiative. Educational Evaluation and Policy Analysis, 29(2), 95-114.
Hill, H. C., Rowan, B., & Ball, D. L. (2005). Effects of teachers mathematical knowledge for teaching on student achievement. American Educational Research Journal, 42(2), 371-406.
Jackson, C. K. (2009). Student demographics, teacher sorting, and teacher quality:
Evidence from the end of school desegregation. The Journal of Labor Economics, 27(2), 213-256.
Kang, N., & Hong, M. (2008). Achieving excellence in teacher workforce and equity in learning opportunities in South Korea. Educational Researcher, 37(4), 200-207.
Klassen, R. M. (2004). Optimism and realism: A review of self-efficacy from a cross-cultural perspective. International Journal of Psychology, 39(3), 205-230.
Lankford, H., Loeb, S., & Wyckoff, J. (2002). Teacher sorting and the plight of urban schools: A descriptive analysis. Educational Evaluation and Policy Analysis, 24(1), 37-62.
Luschei, T. F. (2012). In search of good teachers: Patterns of teacher quality in two Mexican states. Comparative Education Review, 56(1), 69-97.
Luschei, T. F. & Carnoy, M. (2010). Educational production and the distribution of teachers in Uruguay. International Journal of Educational Development, 30(2), 169-181.
Luschei, T. F. & Chudgar, A. (2011). Teachers, student achievement, and national income: A cross-national examination of relationships and interactions. Prospects, 41(4), 507-533.
Martin, C. (1993). The shadow economy of local school management in contemporary
west Mexico. Bulletin of Latin American Research, 12(2), 171-188.
Monk, D. H. (1994). Subject area preparation of secondary mathematics and science
teachers and student achievement. Economics of Education Review, 13(2), 125145.
Monk, D. H., & King, J. A. (1994). Multilevel teacher resource effects in pupil
performance in secondary mathematics and science: The case of teacher subject matter preparation. In R. G. Ehrenberg (Ed.), Choices and consequences: Contemporary policy issues in education (pp. 2958). Ithaca, NY: ILR Press.
Nye, B., Konstantopoulos, S., & Hedges, L. V. (2004). How large are teacher effects?
Educational Evaluation and Policy Analysis, 26(3), 237-257.
OECD. (2005). Teachers matter: Attracting, developing, and retaining effective teachers. Paris: OECD.
OECD. (2009). Creating effective teaching and learning environments: First results from TALIS. Paris: OECD.
OECD. (2010a). Education at a glance 2010: OECD indicators. Paris: OECD.
OECD. (2010b). PISA 2009 results: What students know and can do Student performance in reading, mathematics and science (Volume I). Paris: OECD.
OECD. (2010c). PISA 2009 results: Overcoming social background Equity in learning opportunities and outcomes (Volume II). Paris: OECD.
OECD. (2010d). TALIS 2008 technical report. Paris: OECD.
OECD. (2010e). TALIS user guide for the international database. Paris: OECD.
OECD. (2011). Members and partners. Retrieved from http://www.oecd.org/document/25/0,3746,en_36734052_36761800_36999961_1_1_1_1,00.html
Rivkin, S. G., Hanushek, E. A., & Kain, J. F. (2005). Teachers, schools, and academic
achievement. Econometrica, 73(2), 417-458.
Secretaría de Educación Pública [SEP]. (2011). Alianza por la calidad de la educación. Retrieved from http://www.sep.gob.mx/es/sep1/alianzaporlacalidaddelaeducacion
Tatto, M. T. (2008). Teacher policy: A framework for comparative analysis. Prospects, 38, 487-508.
Wayne, A. J., & Youngs, P. (2003). Teacher characteristics and student achievement gains: A review. Review of Educational Research, 73(1), 89122.
Woessmann, L. (2003). Schooling resources, educational institutions and student performance: The international evidence. Oxford Bulletin of Economics and Statistics, 65(2), 117-170.
World Bank. (2011). World development indicators. Retrieved from http://data.worldbank.org/data-catalog/world-development-indicators