Performing Department
Applied Economics
Non Technical Summary
The question of how rural schools differ from those in denser urban and suburban areas has elicited broad interest from scholars over the years. Much of the existing literature, however, has focused on school finance, economies of scale/size and other ways to deal with population sparsity (e.g. school size/collaboration/consolidations, use of distance education, charter schools, etc.). Some attention has also been directed to rural teacher markets, but the focus has almost exclusively been on differences in pay and teacher credentials (e.g. education, licensure, etc.). This project will examine how several emerging topics of national importance apply in rural settings. These topics include: teacher performance measurement; teacher recruitment, selection and retention; and teacher pay for performance. The project will use data from several sources, including a panel of teacher personnel data maintained by the Minnesota Department of Education, public data on student achievement in standardized test scores in Minnesota, proprietary personnel and applicant data from several school districts in Minnesota, public data on school finance referendum ballots, an education policy database for Minnesota created by the PI, and survey data from the American Community Survey (U.S. Census Bureau).
Animal Health Component
0%
Research Effort Categories
Basic
10%
Applied
80%
Developmental
10%
Goals / Objectives
Human capital is a key prerequisite to economic success and it has historically been relatively scarce in rural U.S. communities. Despite major improvements over time, a sizable rural-urban gap in education among adults has continued to persist over the last four decades, albeit with divergent trends across different levels of education. While the rural-urban gap in adults with no high school degree has been closing, the gap in adult college graduates has widened. The persistence of such gaps in education among adults, combined with the fact that remoteness and population sparsity can create unique obstacles for rural schools, has led to a substantial body of research on the performance of rural schools. Much of the existing literature, however, has focused on school finance, economies of scale/size and other ways to deal with population sparsity (e.g. school size/collaboration/consolidations, use of distance education, charter schools, etc). Some attention has also been directed to rural teacher markets, but the focus has, almost explicitly been on differences in pay and teacher credentials (e.g. education, licensure, etc). An increasing body of work in economics of education has now established a few important facts about teachers. These facts change the landscape for k-12 education research and policy: a-Teachers are the most important school-based factor in student success; b-Teacher performance can be measured reliably by using student outcome data and/or peer observations; c-there is large and somewhat persistent variation in effectiveness across teachers; d-this variation is not explained by any of the factors typically observed about teachers, including education, licensure and other credentials. These factors imply that new mechanisms to measure teacher performance and non-traditional ways recruit, retain and reward effective teachers are needed, and a large amount of research on these topics has ensued (e.g. see Jackson 2014 for a review of the literature). Improvements in educational data systems and a growing body of knowledge on teacher effectiveness has led to a recent surge in policies that target the market for teachers, with multibillion dollar federal programs (e.g. Race to the Top and Teacher Incentive Fund), as well as efforts to implement alternative teacher pay or retention mechanisms in the majority of US states and cities (e.g. see Neal, 2013 for a review). The bulk of this effort, however, targets urban areas, and implications of this changing landscape in education research and policy for rural education remains under examined.This project will examine how several emerging topics of national importance apply in a rural setting; teacher performance measurement; teacher recruitment, selection and retention and teacher pay for performance.Formal objectives are to:Examine relative pay differences for potential teachers by degree and occupation in rural vs urban areasDevelop a framework for generating measures of teacher effectiveness that are partially based on student outcomes for all teachers in Minnesota and predictors of effectivenessExamine rural-urban differences in teacher output, credentials and career choicesExamine how pay-for-performance is applied in rural schools, and its effectsTo accomplish these goals, the project will use data from several sources, including a panel of teacher personnel data maintained by the Minnesota Department of Education, public data on student achievement in standardized test scores in Minnesota, proprietary personnel and applicant data from several school districts in Minnesota, public data on school finance referendum ballots, an education policy database for Minnesota created by the PI, and survey data from the American Community Survey (U.S. Census Bureau).
Project Methods
Goal 1.This study will use data from the American Community Survey on highest educational degree and field of study for individuals interviewed between 2009 and 2014. Lifetime earnings profiles by degree and occupation similar to those presented in table 2.1 will be computed for urban and rural areas separately. Lifetime earnings profiles are generally built by using the earnings of different cohorts of individuals (identified by birth-year) within each survey year to build a schedule of returns to experience within each degree, major and occupation. Using multiple years of survey data allows for removing any year-specific market based effects (e.g. booms, recessions, periods of rapid change in certain occupations, etc.).To identify rural areas, ERS rural-urban continuum codes will be used. These codes are defined at the county level, but county is not always reported in the ACS. Instead, Public Use Microdata Areas (PUMAs) are reported for all records. In dense areas, a county may contain several PUMAs (thus county level rural-urban codes can be applied). In sparser areas PUMAs can be larger than counties, making it impossible to pinpoint the county of residence in each person each year. ACS 3 and 5 year aggregate data do, however, report more geographic detail. With annual data the average rural-urban code across counties contained in a puma can be used, while the multiple year aggregate files are able to pinpoint county, but with a somewhat more limited ability to account for short-term variation in wages. These tradeoffs will be considered and robustness to choice of geography and time will be examined.Career wage paths by major, occupation and rurality can be compared in terms of net present values (NPVs) over careers, first 5 or 10 years.Goals 2 & 3I will use data from Minnesota to pursue objective 2. Sources include: statewide personnel data on all public school teachers, position histories, compensation, licensure, demographics, etc; statewide data on student achievement in standardized scores (public); personnel, applicant and student achievement data from at least one collaborating school district. To measure teacher effectiveness two models will be used, within district and across.The within-district VA model for each grade and subject is: TestScorei=β0+β1TestScoreij+β2 Xi +∑αtTeacheri+εi. That is, current test scores for student i are regressed on lagged test scores for the same student i in all tested subjects j, on a vector of student characteristics Xi and on teacher fixed effects for each teacher t. αt represents the portion of test score growth among a teacher's students attributable to each teacher. The vector X includes indicators for free or reduced price lunch eligibility, english learner status, special education status, mobility, gender, race and ethnicity. The within district model is only estimated in collaborating districts because each district is able to generate roster variables that create links between teachers and students exposed to each other.With the statewide data, students and teachers can only be linked at the school-grade-year-subject level. The state's personnel database has position assignment codes that pinpoint each teacher's assignment in terms of school, grade taught, subject and FTE equivalent for each assignment in each year. This can be used to tie small clusters of teachers to students in most cases. Students and teachers can be linked uniquely in small schools where there is only one teacher serving many school-grade-year-subject level cohorts of students, which is the case in many rural areas. In larger schools, where more than one teacher could be assigned to the same school-grade-year-subject, growth can only be attributed to the cluster of two teachers. It can be further attributed to individual teachers if there is sufficient movement of teachers across schools and assignments over time.Within district performance data and applicant data from at least one district will be used to identify pre-market predictors of performance. Across district (statewide) data will be used to examine differences in performance and career paths across rural and urban areas in Minnesota. Previous literature has addressed questions such as whether teachers are more likely to transition from one type of area to another, and have compared credentials of teachers across urban and rural areas. This project will examine if the relationship between effectiveness and career transitions is different in the sparser rural areas. For instance, there is some recent evidence that teachers that exit the profession are, on average, more effective than incumbents (e.g. Wiswall, 2013). Presumably this is because those who teach more effectively may be more effective in other occupations as well and are drawn away by higher pay in almost all other occupations requiring similar levels of education. In sparse rural areas, if competition from private sector is thin, effective workers may have fewer opportunities to exit. Alternatively, effectiveness may also be related to job satisfaction. If one feels they are in the position that they want to be, one might be less likely to burn out. In thicker markets with more choice, workers in every occupation might be more likely to be there 'by choice' rather than necessity.Goal 4In addition to student achievement data, a database of program rules and participants in Minnesota's Q-Comp program will be used. Q-Comp offers districts in Minnesota up to $260/pupil/year to implement reforms that include Pay for Performance (P4P) schemes. The Minnesota Department of Education (M.D.E.) administers Q-Comp, regulating the range of proposed district operations that are fundable, types of P4P contracts, etc., that would qualify. Proposals have to at least include P4P based on individual teacher, measureable outcomes (although not necessarily on value added per se), on school or grade-level test score growth (a group bonus) and on peer evaluations using a standardized rubric. In addition, districts would implement other changes related to teacher professional development. The district would choose whether to apply to Q-Comp, and how to allocate funds between P4P and other programmatic changes, and how much stakes to tie to each kind of P4P. The state would whether to accept or reject each applying district's proposal. A few participating districts subsequently dropped out of Q-Comp. Therefore, in each year, each district has either not applied, been rejected, has adopted, or has dropped Q-Comp. Different districts applied for and adopted the program each year. Each district's Q-Comp history is coded based on archives of districts' applications, letters from the state granting/denying approval, and district progress reports provided to the PI by the M.D.E.The fact that various districts applied and adopted/got rejected in different years allows for separating the impact of intent to pursue reform from actual reform, and estimate district and year effects separately from program effects. The same generalized differences in differences approach to estimate effects of P4P on student achievement will be taken as in Sojourner, Mykerezi and West (2014). As noted, Q-Comp is one of very few large and long lasting teacher labor market reform programs to make a significant effort to reach rural areas. In fact, M.D.E produced special educational materials and provided supplemental support to encourage Q-Comp adoption in rural Minnesota. Appendix B. presents an example of such material, including a more detailed description of the components of Q-Comp. Examining predictors and differences of program participation, design and impact across rural and urban areas can shed unique insight into the potential uses and effects of P4P in rural K-12 education.