The Effects of Teacher Experience on Budget

Allocation and Student-Teacher Ratios:

An analysis of the affects of teacher experience on student teacher ratios and
budget allocation in Oregon School Districts

Chris Deal
Matthew Dodier
Jing Li
Chong Zhao

Presented to the Department of Economics at the University of Oregon,

in partial fulfillment of the requirements for honors in Economics.

Under the supervision of

Professor Joe Stone

June 2015

Abstract

The economic literature exploring the potential negative externalities of teacher experience is

minimal. To improve on that end we analyze the relationship between teacher experience and

both its effect on marginalized categories of the school budget and on student-teacher ratios at

the district level. In our analysis, we use a panel data set provided by the Oregon Department of

Education spanning the years 2000-2014 across approximately 200 Oregon public school

districts. We utilize logarithmic variations of the Ordinary Least Squares empirical models that

control for socioeconomic and demographic differences, and include district and year fixed

1
effects. Our estimates suggest that an increase in teacher experience leads to a decrease in the

percentage of the budget allocated to athletics, the arts, physical education, and extracurricular

activities. The effects of the average level of teacher experience on physical education, athletics,

and extracurricular activity spending are significant at the five-percent level, while the effect on

the arts is significant at the ten-percent level. Furthermore, our results indicate that there is a

positive relationship between years of teacher experience and student-teacher ratios, an effect

that is statistically significant at the one-percent level.

Acknowledgements: We are grateful to Professor Joe Stone for his patience and helpful

guidance throughout the project, and also to Brian Reeder and Chelsea Clinton at the Oregon

Department of Education for their answers and help with providing the data. We would also like

to extend a much appreciated thank you to Josh Olsen for his countless hours of patience and

hard work in assisting us with our data problems and regressions.

2
 Table of Contents
I. Executive Summary…………………………………....Page 4

II. Introduction…………………………………………….Page 6

III. Literature Review………………………………………Page 9

IV. Data Description……………………………………….Page 12

V. Hypothesis & Methodology…………………………...Page 16

VI. Results……………………………………………….…Page 20

VII. Discussion of Results……………………………..……Page 24

VIII. Conclusion………………………………...……………Page 26

IX. References…………………………………..………….Page 28

X. Appendix…………………………………………….…Page 31

3
 I. Executive Summary

Many current concerns facing public schools in Oregon include low on-time graduation

rates, the slow growth in average spending per student, and large class sizes. In the wake of

these concerns, a great interest has developed among education researchers and economic

analysts to provide better information on these subjects for future policy recommendations. One

avenue in economic research is to focus on teacher experience as the variable of interest because

it has great policy implications for education researchers, school officials, the general public, and

teachers themselves. Teacher experience has many direct positive effects on student

achievement. Additionally, teacher experience has potential negative effects on both the

allocation to marginalized categories of the school budget and the size of student-teacher ratios,

due to the increased costs associated with employing teachers with a greater amount of

experience. The current research looking at the potential externalities of teacher experience is

very minimal. To improve upon this, we analyze the relationship between teacher experience on

the marginalized budget categories specifically athletics, physical education, the arts, and

extracurricular activities, as well as its effect on student-teacher ratios, all conducted at the

district level.

To accomplish this, we utilize a panel data set provided by the Oregon Department of

Education, which spans the years 2000-2014, across approximately 200 Oregon public school

districts. We employ a logarithmic variation of the Ordinary Least Squares Regression approach

that utilizes integrated fixed effects to control for year and district differences, as well as a vector

of variables that control for both the socioeconomic and demographic differences among the

various districts. Our estimates suggest that an increase in teacher experience leads to a decrease

4
in the percentage of the total district budget allocated to athletics, the arts, physical education,

and extracurricular activities. The effects of the average level of teacher experience on physical

education, athletics, and extracurricular activity spending are significant at the five-percent level,

while the effect on the arts is significant at the ten-percent level. Furthermore, our results

indicate that there is a positive relationship between years of teacher experience and student-

teacher ratios, an effect that is statistically significant at the one-percent level.

Based on the findings we believe that changes to teacher experience have a strong impact

on marginalized categories of the budget and also on district level student-teacher ratios.

However, further research is needed to understand the cost tradeoffs and mechanisms associated

with increases to years of teaching experience. Future avenues of research could analyze more

deeply the mechanisms of student-teacher ratios at the district level, specifically how an increase

to the student-teacher ratio changes the composition of the faculty and administrative staff.

Additionally, researchers could trace the path of cost effects from teacher experience to salary,

pensions, and number of employed teachers and how these same costs and tradeoffs affect

student achievement outcomes. These approaches may yield greater understanding of the

mechanisms of which these externalities work and also more awareness of how funding

decisions are made.

5
 II. Introduction

Public Education in Oregon is of interest to policy makers, social scientists, and the

general public in order to increase opportunities and outcomes for students and because it greatly

impacts the state economy. However, there are many present concerns in Oregon public school

districts. For example the U.S. Department of Education showed that Oregon's class of 2013

graduation rate ranked worst among the 49 states measured, while it ranked 5th worst for the

class of 2013 under the new definition of on-time graduation rate. 1 Notably, the on-time

graduation rate reached a historic level of 72 percent for the class of 2014, but this can be

attributed due to the new definition.2 Another concern for education officials in Oregon are the

implications of large class sizes on students, because Oregon’s public school system has some of

the largest class sizes in the nation.3 In this paper we look specifically at student-teacher ratios,

which are mathematically related, but distinctly different from class size.4 Regarding student-

teacher ratios, in 2011-2012 Oregon had 21.2 students for every teacher, compared with 16

students per teacher nationwide. This ultimately means that the average Oregon teacher

shouldered 33 percent more students than the average U.S. teacher, which has been a growing

trend since the 1990s (Hammond 2014a).5 Furthermore, policy makers are concerned with the

implications of teacher cutbacks and shorter school years, in addition to slow growth in average

spending levels per student. For example, the first time the Census Bureau recorded how much

states were spending per student, in 1957, Oregon was found to be the highest spender in the

1
The definition of on-time graduation now includes special education students who earn modified diplomas, as well
as students who delay graduation to remain in high school to attend community college at public expense
2
See Hammond 2015 for more information recent graduation rates in Oregon
3
No agency requires Oregon middle schools or high schools to report the number of students in every class, so there
is no uniform definition or specific mechanism of how schools should define a class or the number of students in it.
4
Student-teacher ratios are defined as the number of students per full-time equivalent teacher
5
The ratio is often used as a proxy for class size. In most cases, the student–teacher ratio will be significantly lower
than the average class size, with class sizes commonly being 30% or more than student-teacher ratios

6
nation, however Oregon’s ranking has greatly changed over the last 50 years. Unadjusted

spending per student in the years 2011-12 was 11 percent lower than the national average, which

is the biggest gap that Oregon has ever trailed the national average. 6 In the wake of these

concerns there is great interest by education researchers to provide better information for future

policy recommendations and to better understand the long-term trends. One avenue in economic

literature on education is looking at the relationship between budget composition and student

outcomes. This track of research is often limited to studying student achievement based on the

allocation of the budget to certain programs that affect their educational outcomes. Another

avenue for research is exploring the role of class size, or student-teacher ratios and their impact

on student achievement, test scores, and educational outcomes.

Few studies have looked at budget allocation or student-teacher ratios from a perspective

other than their relationships with student achievement. In recent years policy makers and the

general public have debated the effects of teacher quality and experience and its role in shaping

student outcomes. This leaves teacher experience as an important variable of interest with policy

implications for education researchers, school officials, teachers and the general public. Teacher

experience has many direct and indirect effects on both budget composition and student

achievement. For example, more experienced teachers have positive direct effects on student

achievement in the classroom, due to the fact more experienced teachers are better equipped to

deal with and educate students. This is especially true when teaching students that come from

socially disadvantaged backgrounds. More experienced teachers also typically have more

knowledge and expertise than less experienced educators. Teacher experience also has potential

negative indirect effects on particular elements of the budget, such as music, arts, drama, and

6
For more information about average spending per student in Oregon see Hammond 2012 and 2014b

7
sports, because less money can be allocated to those categories when incurring higher costs

associated with more experienced teachers.

Our main estimates find that teacher experience has a negative effect on many different

categories of school district budgets. The effect of average teacher experience on athletics

(sports), physical education and extracurricular activities are all statistically significant at the five

percent level. Further, teacher experience negatively effects the allocation of the budget to the

arts, an effect which is significant at the ten-percent level.

In our principal findings concerning the effect of teacher experience on student-teacher

ratios we find that the coefficient for average teacher experience is positive and is approximately

0.102. Due to the logarithmic specification of our model, this means that for every ten percent

increase in average teacher experience in a school district, there is an increase in the student-

teacher ratio in by 1.02 percent.7 This effect is significant at the one-percent significance level

after controlling for heteroskedasticity by employing robust standard errors. These results

indicate that school districts that employ teachers with a higher average level of teacher

experience ultimately have higher student-teacher ratios. This is because they must allocate a

larger portion of their budget to salary and pensions costs and consequently are unable to afford

as many teachers as school districts that employ teachers with lower levels of experience.

The paper proceeds as follows. Section III reviews relevant economic literature

concerning school budget composition, teacher experience, student-teacher ratios and education.

Section IV reviews the data sources used in the estimation and section V reviews the hypothesis

and methodology of our paper. Section VI presents the main estimates, Section VII discusses the

results, and Section VIII concludes.

7
Our model uses logged independent variables as well as logged dependent variables. For more information see the
Hypothesis & Methodology section, as well as the Results section

8
 III. Literature Review

Although few economic papers have explored the relationship between district budget

composition and academic achievement, many have explored the links between class size and

student outcomes. Additionally any relevant research utilizing teacher experience as the key

variable of interest is minimal. 8 Though existing research focusing on teacher experience is

minimal, it is important to note that prior literature regarding education policy analysis has

provided some basis for the work that follows in the rest of the paper.

Theoretically, allocation to specific categories of the budget can potentially be affected

by the average level of teaching experience in a district because more experienced teachers earn

higher salaries and larger pensions, which decreases the amount of the budget that is available to

allocate to other areas. Prior economics literature concerning school budgets has concentrated on

the effectiveness of the various different inputs on student performance. Many studies have

looked at the link between school funding and student performance, but these papers did not look

at the relationship between teacher experience and how school funding is allocated.9 However,

a previous study by Stephen M. Barro and Stephen J. Carroll finds that the level of budget

spending on teachers can affect the budget allocation in variety of different aspects. First, an

increase in budget spending on teachers, either by hiring more teachers or increasing the overall

pay schedule for teachers, is generally a high priority for education officials and policy makers

when extra funding becomes available. Barro and Carroll (1975) find that budget spending on

teachers will increase by approximately 0.75 percent when the total budget available for the

district increases by 1 percent. Essentially, when additional funds are available, teachers benefit

8
We look at student-teacher ratios in our paper, but most economic papers look at or use the term class size
9
To find papers examining the effects of school funding on student performance see Hanushek (1981, 1986, and
1989)

9
by having more of the budget allocated towards employment costs such as salary and pensions.

However, the role of experience is not explicitly mentioned in this paper and we explore the

potential relationship of teaching experience to improve this previous analysis.

Further, the research involving student-teacher ratios, class sizes and the research on

teachers is robust, but it has often focused on its effects on student achievement. Traditionally,

the research on student-teacher ratios or class size has analyzed large datasets containing

information on a variety of educational inputs and student outcomes (Rockoff 2009).10 Recently,

economists from the past few decades have used experimental and quasi-experimental variation

to build stronger casual relationships and account for unobservable factors. 11 However, there is

not any literature that directly looks at the role of teacher experience in increasing class sizes.

Additionally, the literature on teachers is also vast and covers a wide range of topics

such as looking at the effects of teacher effectiveness and other teacher characteristics on student

achievement.12 Yet, this literature mostly looks at teacher quality, certifications (see Kane,

Rockoff, and Staiger 2008), and other related qualifications, and whether or not these additional

qualifications are correlated with stronger student performance and more effective teaching. 13

Essentially, the studies that we are aware of only looked at different attributes of teachers as a

measure of their effectiveness rather than looking at potential effects on student-teacher ratios or

different categories of the budget.

10
See Hanusek (1998) and Krueger (2003) for two different overview and analyses of the vast amount of literature
on class size
11
For information see Angrist and Lavy (1999), Krueger (1999), Hoxby (2000). See Rockoff (2009) for an
overview of field experiments on class size in the early 20 th century
12
See Hanushek (1971), Murnane (1975), Ferguson and Ladd (1996), Rockoff (2004), Rivkin, Hanushek, and Kain
(2005), and Angrist and Guryan (2008).
13
Murane (1975), Summers and Wolfe (1977), Ehrenberg and Brewer (1994), Clotfelter, Ladd, and Vigdor (2006)
and Aaronson, Barrow, and Sander (2007) have failed to consistently find that teachers holding master’s degrees are
more effective

10
 However, a literature outside of economics finds that state decision makers tend to use

additional local funds to employ teachers with more professional teaching experience (Macphail-

Wilcox and King, 1986). Both of the observations imply the budget composition will be affected

by teacher salary and pension costs, which is determined primarily by years of teaching

experience. In terms of the second track of our paper, which is to analyze the relationship

between teacher experience and student-teacher ratios, Macphail-Wilcox and King find that

teachers with higher salaries are mainly employed in districts that receive higher funds and have

lower levels of minority students. Furthermore the authors find that student-teacher ratios are

lower in schools where the percentage of minority student in enrolled is higher. This implies that

the schools with more minority students hire more teachers. Since teaching salary is tied to

experience these results imply that teachers experience is leading to higher teacher costs and

potentially larger student-teacher ratios and class sizes. In our paper we attempt to examine this

mechanism directly to draw stronger conclusions about the relationship between experience and

student-teacher ratios. Drawing from the previously stated literature, we assume that the

student-teacher ratio will be lower in the school districts that have more minority students

enrolled. Additionally, to control for the effects of minority students on budget allocation and

student-teacher ratios we include race dummy variables and a variable for the percentage of

students in free and reduced lunch programs in our regression models.14

Although there is research that explores mechanisms of education funding and budget

composition in addition to papers on student-teacher ratios no study that we are aware of has

attempted to measure the effect of teacher experience on either budget allocation or student-

teacher ratios. Much of the role of teaching experience is implied in these studies, but relatively

14
For more details about these control variables see the regression equations in the Hypothesis & Methodology
section of the paper

11
little is known about it as a key indicator variable and its resulting magnitude of effect. By

measuring the average level of teacher experience in school districts, we hope to capture the

casual effect of teacher experience on both budget allocation in our first track of research and its

impact on student-teacher ratios in the second track of our study.

IV. Data Description

The data sources we use in our analysis are from the Oregon Department of Education’s

annual reports on teachers, student demographics, and school district budgets. 15 There over three

million observations collected by the Oregon Department of Education (ODE) which are

organized into a panel dataset and comprise observations from around 200 Oregon public school

districts across a time period of 15 years. Based on this we hypothetically have around 3000

observations on the dependent variables, which are the different categories of the school district

budget and student-teacher ratio respectively, with the unit of analysis at the school district-

level.16 The teacher data includes observational variables such as the number of employed

teachers and average years of teaching experience within a district. Examples of student level

variables include the total student enrollment, demographic characteristics of the students and

percentage of students in free and reduced lunch programs.17 Furthermore, the data on school

district budgets has observations for various budget categories such as spending on general

classroom instruction, social studies, and health education. However, in our paper we restrict our

analysis to specifically the following budget categories: the arts, athletics, physical education

(P.E.), and extracurricular activities.

15
To see the variables we use in our regressions see Table 1 in the appendix
16
These categories are athletics, physical education, the arts, and extracurricular activities
17
See Table 1 to see other student characteristics we use in our regressions

12
 Figure 1 below illustrates the average percentage of budget allocation for the

marginalized categories, specifically the arts, athletics, physical education and extracurricular

activities across the 15 school years. Even though the marginalized items, as shown below, only

occupy small portion of the total school budget, we determine based on our methodology that

teacher experience has statistically significant effect on the percentage change of budget

allocations for these marginalized items.18

FIGURE 1:

Average Yearly Budget Allocation for Marginalized

Items in Oregon Public School Districts between School
Years 2000-2014

The Arts 3.24

Physical Education 2.11
Other Extra-curricular Activities 1.47
Athletics 1.63

0 0.5 1 1.5 2 2.5 3 3.5

Budget allocation for marginalized items, in percentage

For teacher statistics, the data collected by Oregon Department of Education shows that

the average number of teachers reached a peak in the 2007 school year and was followed by a

large decrease until the 2012 school year, as indicated by Figure 2 on the following page.19

When evaluating the trend of average teacher experience across the 15 school years, shown in

Figure 2, we notice that teaching experience surprisingly increased while many teachers left

school and pursued other career options. Between school years 2005 and 2006, the average

18
See Results section for further information about our findings and significance levels and see Table 2 in the
appendix for our Final Regression Results
19
We utilize years of teaching to represent teacher experience, our key variable of interest.

13
number of teachers employed by school districts in Oregon increased by only two whereas

teacher experience reduced by over eight months on average. Further, the inverse correlation

stopped and briefly became a positive correlation when both teacher experience and net number

of teachers decreased between school years 2007 and 2008. Based on the data, we believe that

many experienced teachers left districts or were laid off during the Great Recession and an

increasing amount of less experienced teachers were employed by the Oregon public school

system around the 2008 academic school year. Additionally, beginning with the 2009 school

year the inverse correlation between teacher experience and number of teachers in district

returned and continued for the following years in our data.

FIGURE 2:

Summary for Teachers in Oregon Public School

155 Districts, School Years 2000-2014 9.8

9.6
150
9.4
Avg. Number of Teachers

Years of Teaching
145 9.2

9
140
8.8

135 8.6

8.4
130
8.2

125 8
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Teacher Experience Num. of Teachers 2 per. Mov. Avg. (Teacher Experience)

Notably, in 2012 the average number of teachers was the lowest across Oregon public

school districts in the 15 years of data that was analyzed. Further, the average teacher experience

14
at the school district level was at its second highest in this same year, and we believe this may

correlated with the low number of employed teachers for that specific year.20 Starting from

school year 2012, on average the net number of employed teachers increased yearly by three in

Oregon public school districts while the average years of teaching experience decreased annually

by more than three months.

As for the total yearly budget assigned to Oregon school districts, it follows an overall

smooth trend as the blue bars indicate in Figure 3 below. To find the effect that teacher

experience has on budget allocation, we evaluate the trend of teacher experience versus total

district yearly budget across the past school years. The total budget fluctuated overall

throughout the data, whereas average years of teaching continued to decrease until school year

2008. After that, average years of teaching in districts experienced an increase until school year

2010. In addition, the total district yearly budget rose between school years 2010 and 2012,

while the average teacher experience was stable before it increased.

FIGURE 3:

Total District Year Budget and Teacher Experience in

Oregon Public School Districts, School Years 1999-
2012
Avg. Total Annual Budget, in $10,000,000

3.5 10
3
Years of Teaching

2.5 9.5
2
9
1.5
1 8.5
0.5
0 8
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
School Year
Total District Year Bgt. Years of Teaching

20
The highest level of average teacher experience was in 2002.

15
 The graph in Figure 4 exhibits the trend of student-teacher ratios in Oregon public school

districts and shows the average teacher experience across 10 school years. Overall, there is a

positive correlation between the variables shown below. As student-teacher ratios decreased and

reached its lowest level in the 2007 school year, years of teaching also consistently decreased

over the same period. Moreover, both average years of teaching and average teacher experience

started to increase after the 2008 school year until the 2012 school year. The most recent data in

the figure shows a decrease in average teaching experience while student-teacher ratios remained

stable.

FIGURE 4:

Summary for Teacher Experience and Student-Teacher

Ratios in Oregon Public School Districts, School Years
2004-2013

Years of Teaching
15.5 10
Student-teacher Ratio

15 9.5
14.5
9
14
13.5 8.5

13 8
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013
School Year

STR Years of Teaching

V. Hypothesis & Methodology

The premise for our paper is to measure the potential negative externalities resulting from

increases to the average level of teacher experience. Teaching experience has positive direct

effects on student achievement within the classroom, but at the same time has indirect negative

16
effects. Our hypothesis is that these indirect negative effects are specifically seen in two

different areas, first in the decreasing allocation of funding to budget categories like the arts, and

athletic spending due to the increasing pension and salary costs in the budget and secondly in the

increasing student-teacher ratios due to the higher costs associated with more experienced

teachers. A public school teacher’s salary is comprised of two components that are directly

affected by experience; their pension and their yearly salary. Each component of a teacher’s

salary increases in tiers as their years of experience increase. This ultimately results in more

experienced teachers costing more to employ both from a pension and salary standpoint. The

additional costs associated with these teachers causes funding to be drawn from other budget

items. Essentially, it is our expectation that the additional budget expenditures associated with

the increases in teacher’s salaries and pensions will result in lower allocation to marginalized

budget items such as the arts and athletic spending in addition to larger student-teacher ratios.21

In an attempt to measure the outcomes of our key independent variable, years of teaching

experience, we split our methodology into two distinct paths. The first path is to examine the

relationship between teaching experience and allocation to specific marginalized categories of

the budget, while the second path is an examination of the casual relationship between increases

in teaching experience and student-teacher ratios. To do this we use a panel data set provided by

the Oregon Department of Education which consists of around 200 public school districts in

Oregon covering a 15 year period. We employ a logarithmic specification of the Ordinary Least

Squares Regression approach in our analysis as well as integrated fixed effects in order to control

for year-to-year and district-to-district fluctuations. In addition, we add controls to take into

account the different socioeconomic status of individual school districts. Examples of variables

that are controlled for include total budget, total school enrollment, and percentage of students on
21
Since student-teacher ratios are mathematically related to class sizes this will also result in larger class sizes

17
free and reduced lunch programs. Here, the enrollment and free lunch program variables are

used as controls for the effects of student population and low-income status respectively.

Presented below is further analysis of the methodology employed in our analysis of the potential

externalities of teacher experience.

4.1- Effect of Teacher Experience on Budget Categories

Controlling for the differences across years and across individual school districts that

may affect a district’s overall level of budget allocations, we isolate the effect that teacher

experience has on budget allocations by using an integrated fixed effect model for the

marginalized items of the school district budget. The marginalized budget areas focused upon are

funding to physical education, athletics, the arts (i.e. drama, music, and theater), and

extracurricular activity programs.

To test our primary hypothesis that teacher experience is associated with a reduction in

funding allocation to marginalized items of the budget because of the increased costs of

employing more experienced teachers, we utilized the presented regression specification below.

Here, we use an integrated fixed effect model to control for the differences across years

and across districts and evaluate the effects of teacher experience on budget allocations (i.e. the

arts, athletic spending, P.E., and extracurricular activities). The intercept term, is a vector of

fixed effects that will be controlled for in the model. The variable is the logarithm of budget

category j for district i in year t. The fixed effects that are being controlled for are the

differences across years and across individual school districts. The variable is the logarithm

of the average teacher experience in district i for year t. The measure of teacher experience is the

average years of teaching experience across teachers at the school district level. Finally, the

18
variable in the regression is a vector of all the control variables that are utilized in the

regression for a given year and school district. We have the expectation that the coefficient on

teacher experience will have a negative value for marginalized categories of the budget such as

the arts because we expect be the increased costs associated with employing teachers with more

experience will cause funding to taken away from budget items that are susceptible to budget

cuts. The full equation used in one of our regression is as follows:

4.1.1. The effect of Teacher Experience on the Budget Allocation for Physical Education:

Our endogenous variable, , is the logarithm of budget allocation for physical

education for district i in year t. is the logarithm of average teacher

experience in district i in year t. is the logarithm of the total district

budget for district i in year t. Here, and

are controls for student population, income and poverty. As mentioned above, the variables of

race, such as , , , , , and

are the controls for the demographic characteristics of the students in the school

19
districts. is the constant term and is the error term. The coefficient estimates the effect

that teacher experience has on the percentage change of budget allocation for P.E.22

4.2- Effect of Teacher Experience on Student Teacher Ratios

In our second track of research we look at the effect of years of teacher experience on

student-teacher ratios at the school district level, and our regression takes the following form:

In this regression model all independent variables are specified in the same manner as the

previous path, but here the dependent variable is now the logarithm of the student-teacher ratio

for district i in year t.23 Our belief is that districts that have teachers with higher experience will

be forced to allocate a larger portion of their budget to salaries and pensions and thus will not be

able to afford as many teachers as other schools with lower experienced teachers, resulting in

larger student-teacher ratios for the former types of school districts. Furthermore, our

expectation is that the coefficient of the variable for the average years of teaching experience

will have a positive sign, because school districts that employ more experienced teachers will

have less money to employ additional teachers.

VI. Results

In our first track of research we examine the relationship between teacher experience and

budget allocation to physical education spending, athletics spending, spending on the arts and

22
See the appendix for the other specific equation that look utilize the categories of the budget as the dependent
variable
23
See the appendix for the specific equation used in our student-teacher ratio regression

20
extracurricular activities spending.24 Our regression results suggest that at the five-percent

significance level, teacher experience has a negative effect on physical education, athletics and

extracurricular activities spending. Furthermore, at the ten-percent level teacher experience has a

negative effect on spending on the arts. In our regression we use logged dependent variables for

each budget category and logged independent variables. The coefficients for the log of average

level of teacher’s experience are -0.404, -0.271, -0.276 and -0.431 for budget allocated to

physical education, athletics, art and extracurricular activities respectively. In terms of

interpreting our regression results, the coefficient of the log of average level of teacher’s

experience for example is -0.404 in regression for the log of physical education spending. This

result is an elasticity meaning that a one-percent increase in average teacher experience leads to

0.404 percent decrease in the total budget being allocated to physical education spending. To

control for demographic characteristics we use control variables for race, which vary in whether

they have either positive or negative effects on categories of the budget and in their significance

levels, with some being significant and most being highly insignificant. 25 We also use another

socioeconomic status control variable, the variable for percentage of students on free and

reduced lunch at the school district level; this variable is highly insignificant in all regressions

except in the regression looking at spending on extracurricular activities where it is significant at

the five-percent level.26

We also include controls for total yearly district budget and total number of students at

the school district level, both of which are in logarithmic form. Interestingly, the total yearly

district budget is highly significant in only two of our regressions when looking at the effects of

24
All these dependent variables are in logged form, see Table 1 for more information about the variables and see the
Table 2 for R2 values and standard errors, as well as significance levels and coefficients
25
See Table 2 for more information, note the coefficients on these variables vary based on the regression. White is
the control group and is left out of the regression equations
26
See Table 2 in the appendix for more information

21
teacher experience on allocation to specific categories of the budget. The total yearly budget is

significant in the regressions for physical education spending and for athletic spending, while it

is highly insignificant in the two other regressions.

In our second track of research we find that a higher level of average teacher experience

at the school district level will result in an increase in student-teacher ratios.27 The regression

results indicate that the coefficient of the logged average years in school district, which

represents the level of average years of teaching experience, is 0.102 and the p-value is 0.006,

which shows that increasing a teacher experiences has a significant effect on the student-teacher-

ratio at the one-percent significance level. This result implies that a ten-percent increase in

average years of teaching experience will result in approximately a one-percent increase in

student-teacher ratios. A ten-percent increase in average years of teaching experience is roughly

comparable to an increase of one year in teaching experience which would lead to an increase in

student-teacher ratios from 20 to 20.2 students on average.28 This supports our original

hypothesis that more experienced teachers will contribute towards an increasing student-teacher

ratios at the school district level. Essentially, school districts are forced to allocate a larger

portion of their budget to salaries and pensions and consequently will not be able to afford as

many teachers as other school districts that employ teachers with lower levels of experience.

However, further mechanisms may be needed to examine this result, particularly at the

individual school level in order to determine more information about the causal effect of teacher

experience on student-teacher ratios.

27
For R2 values, standard errors, or coefficient values not discussed in the Results section see Table 2 in the
appendix
28
See Figure 5 to see the effect of 10% increase of average level of teacher experience at the school district level

22
 We examine further results concerning the control variables used in our empirical

model.29 We find that the coefficient of log total students in districts is 0.861, which is

significant at the one-percent level meaning that school districts with a higher number of total

students will have larger student-teacher ratio than districts with less total students. This positive

effect may be attributable to the fact that the number of teacher in a district does not increase at

the same rate as the number of students. Furthermore, the coefficient of log total district yearly

budget is -0.138 which is significant at the one-percent significance level with a t-statistic of -

4.58, meaning that a 10 percent increase in the total annual budget in districts will lower the

student-teacher ratio by approximately 1.4 percent. This supports our assumption that larger

increases in the total yearly budget will allow schools to hire more teachers and reduce the

student-teacher ratios. Our race control variables show differing results with some variables

having negative effects on student-teacher ratios and others having positive effects on them.30

However, only the Native American control variable is significant at the ten-percent level, which

has a positive effect on student-teacher ratios. In addition, the variable for percentage of student

on free and reduced lunch, which is a control for low-income status, is highly insignificant and is

not significant at the ten-percent level depending on the individual regression.31 It is important to

note that variables such as the race control variables and the variable for free and reduced lunch

programs are merely being used as controls variables to isolate the effect of teacher experience,

and that more investigation must be done in order to determine why these variables are

statistically significant or insignificant.

29
See Table 2 for results not discussed in the Results section
30
White is the control group for the race dummy variables
31
Has p-value of 0.729 and a t-statistic of -0.35

23
 VII. Discussion of Results

This paper aims to analyze this subject to provide educational departments with more

information to construct their budgets in order to maximize programs that benefit student

achievement while also providing satisfactory financial support to educators. We are not aware

of any previous studies that examine the effects the tracks of research we pursue which leaves

our study contributing to this type of analysis. Our results establish that there is a significant

relationship between teacher experience and both budget allocation and student-teacher ratios,

but there are important caveats to our study. This study looks at budget compositions and

student-teacher ratios at the school district level, potentially future literature could examine the

effects of average level of teacher experience at the individual school level, possibly resulting in

stronger implications due to a greater number of observations. Further, due to time and data

constraints we are not able to trace out the path of cost effects from teacher experience to salary,

pensions, and number of teachers. Potentially both of these approaches may yield greater

understanding of the mechanisms of which these externalities work and also more awareness of

how funding decisions are made.

Additionally future papers could purpose a third track of research, which looks more

deeply at the mechanisms of student-teacher ratios at the district level. It could examine how

increased student-teacher ratios change the composition of teachers at the school district level.

This track of analysis could explore whether increases in student-teacher ratios leads to less

special education teachers employed district wide, less music and arts teachers, or decreases in

library personnel. Future work could also explore more aspects of the budget in addition to

categories that we look at in this paper. Furthermore, there are also some limitations to our

24
findings, for example due to Oregon’s biennium education budget structure; these results may

not necessarily be externally valid for other public school districts nationwide. 32

In terms of interpreting our findings, previous economic literature has mixed findings

regarding the effect student teacher ratios and class sizes on classroom achievement.33 Many

researchers have found that there are positive gains from reducing class sizes, particularly for

minority students (Finn and Achilles 1990, Krueger 1999, and Krueger and Whitmore 2001).

While recent field experiments have shown there are positive gains from reduced classroom sizes

(Bain and Achilles 1986, Word et al. 1990, Krueger 1999, and Krueger and Whitmore 2001).

On the other hand, other literature from educational economics suggests that class size has

minimal effects on student achievement (Hanushek 1986, 1998).

Concerning teachers, research has consistently found that teachers are important for

student learning and achievement, but the findings concerning the effectiveness of teachers are

mixed (see Rockoff 2004, Rivkin, Hanushek, and Kain 2005, Aaronson, Barrow and Sander

2007, and Kane, Rockoff, and Staiger 2008).34 Additionally, economics literature examining the

relationship between salaries and measures of teacher quality or performance has a lot of

variation in its findings (see Murnane et al. 1991, Hanushek, Kain, and Rivkin 1999, and Figlio

2002).

However, our research is not to suggest that teachers should be compensated less or that

they are the main cause of increased student-teacher ratios and reduced allocation to

marginalized categories of the budget. Rather our findings suggest that there are externalities

32
Biennial budgeting typically works on an odd-even system. Lawmakers submit and approve a budget that includes
24-month appropriations on an odd year and focuses on budget oversight in even years. Essentially, school districts
on a biennium education budget may make certain adjustments that districts on a single year budget structure would
not due to the type of budget.
33
See Cohn and Millman (1975), Glass and Smith (1978), Hanushek (1986), and Hedges, Laine, and Greenwald
(1994) for more information on the effects of class size on student achievement
34
For a few examples of overviews of the literature discussing teacher effectiveness see Hanushek and Rivkin
(2012) and Jackson, Rockoff, and Staiger (2014)

25
associated with teacher experience that are less understood. They suggest that budgets and

student-teacher ratios are very sensitive to changes in the level of teacher experience and propose

that teacher costs are one of many mechanisms that are resulting in these types of funding

decisions being made. Broadly, our results suggest that further attention by education officials

and policy makers is needed to explore these negative externalities associated with increases in

teacher experience.

VIII. Conclusion

Increasing student-teacher ratios and shrinking budgets continue to be a major concern

for public school systems nationwide. Despite strong public interest, the effects of school

budgets have received little attention from education policy researchers. Outside of examining

the impact of budget composition on student achievement little is known about their impact on

outcomes of interest to policy makers, social scientists, and the general public. The education

literature on student-teacher ratios is narrowly tailored towards exploring its effect on student

achievement and other student outcomes. This study attempts to alleviate those concerns by

using teacher experience as our variable of interest and look at its effect on both specific

categories of school district budgets and student-teacher ratios. Theoretically we expect teacher

experience to have negative effects on particular elements of the budget, such as music, arts,

drama, and sports, while also increasing student-teacher ratios because more experienced

teachers are typically more expensive in terms of salary and pension costs.

Our regression findings suggest that an increase in teacher experience leads to a decrease

in the funding allocated to athletics, arts, physical education, and extracurricular activities. The

26
effect of the average level of teacher experience on athletics, physical education and

extracurricular activities spending are significant at the five-percent level and the effect on

spending on the arts is significant at ten-percent level.

In addition, we also explore the relationship between teacher experience and student-

teacher ratios. We find that an increase in the teacher experience is associated with an increase

in student-teacher ratios, an effect significant at the one-percent significance level. Our findings

are robust to a wide set of specifications, including district and year fixed effects, district-specific

linear time trends, and controls for time-varying factors such as the percentage of students on

free and reduced lunch, total budget, and student enrollment.

With respect to policymakers and education officials, our findings suggest that any

informed debate over budgeting will need to weigh the benefits of increases in teacher

experience against the potential negative externalities that may result.

27
 IX. References

Aaronson, Daniel, Lisa Barrow, and William Sander. 2007. "Teachers and student achievement
in the Chicago public high schools." Journal of Labor Economics 25(1), 95-135.
Angrist, Joshua D., and Jonathan Guryan, 2008. "Does teacher testing raise teacher quality?
Evidence from state certification requirements." Economics of Education Review 27, no.
5: 483-503.
Angrist, Joshua D., and Victor C. Lavy. 1999. “Using Maimonides’ Rule to Estimate the
Effect of Class Size on Scholastic Achievement.” Quarterly Journal of Economics,
114(2): 533–75.
Bain, Helen P., and Charles M. Achilles. 1986. “Interesting Developments on Class Size.” Phi
Delta Kappan, 67(9): 662–65.
Barro, Stephen M., and Stephen J. Carroll. 1975. “Budget allocation by school districts: An
analysis of spending for teachers and other resources.” Rand, 1975.
Clotfelter, Charles T., Helen F. Ladd, and Jacob L. Vigdor. "Teacher-student matching and the
assessment of teacher effectiveness." Journal of Human Resources, 41(4), 778-820.
Cohn, Elchanan and Stephen D. Millman. 1975. “Input-ouput Analysis in Public Education.”
Cambridge, MA: Ballinger.
Ehrenberg, Ronald G., and Dominic J. Brewer. 1994. "Do school and teacher characteristics
matter? Evidence from high school and beyond." Economics of Education Review 13(1),
1-17.
Ferguson, Ronald F. and Helen F. Ladd. 1996. “How and why money matters: An analysis of
Alabama Schools.” In H. Ladd (Ed.), Holding schools accountable (pp. 265–
298).Washington, DC: Brookings Institution.
Figlio, David N. 2002. Can public schools buy better-qualified teachers? Industrial and Labor
Relations Review, 55,686–699.
Finn, Jeremy D., and Charles M. Achilles. 1990. “Answers and Questions About Class Size: A
Statewide Experiment.” American Educational Research Journal 27 (3): 557–77.
Glass, Gene V., and Mary L. Smith. 1978. Meta-analysis of Research on Class Size and
Achievement. San Francisco, CA: Far West Laboratory
Hammond, Betsey. 2012. “Oregon spent more per student on education than any other state in
1957.” OregonLive.com. September 18.
http://www.oregonlive.com/education/index.ssf/2012/09/oregon_spent_more_per_studen
t.html
Hammond, Betsey. 2014a. “Class Size Controversy: Portland Elementary Schools Have Much
Smaller Classes than Nearby Suburbs.” OregonLive.com. February 15.
http://www.oregonlive.com/education/index.ssf/2014/02/class_size_controversy_portlan.
html.
Hammond, Betsey. 2014b. “Education Spending in Oregon Fell for Three Years, Reached
Record Low Compared to National Average, Study Says.” OregonLive.com. May 23.
http://www.oregonlive.com/education/index.ssf/2014/05/education_spending_in_oregon_
f.html.
Hammond, Betsey. 2015. "Oregon fails to improve dismal graduation rate.” OregonLive.com.
January 29.

28
 http://www.oregonlive.com/education/index.ssf/2015/01/oregon_fails_to_improve_disma
l.html.
Hanushek, Eric A. 1971. “Teacher Characteristics and Gains in Student Achievement:
Estimation using Micro Data.” American Economic Review, 61(2): 280–88.
Hanushek, Eric A. 1981. "Throwing money at schools." Journal of policy analysis and
management 1, no. 1: 19-41.
Hanushek, Eric A. 1986. “The Economics of Schooling: Production and Efficiency in Public
Schools.” Journal of Economic Literature, 24(3): 1141–77.
Hanushek, Eric A. 1989. "The impact of differential expenditures on school performance."
Educational researcher 18, no. 4: 45-62.
Hanushek, Eric A. 1998. “The Evidence on Class Size.” Occasional Paper Number 98-1, W.
Allen Wallis Institute of Political Economy, University of Rochester, Rochester, NY,
February.
Hanushek, Eric A., and Steven G. Rivkin. 1997. “Understanding the Twentieth-century
Growth in U.S. School Spending.” The Journal of Human Resources, 32(1): 35–68.
Hanushek, Eric A., and Steven G. Rivkin. 2012. "The distribution of teacher quality and
implications for policy." Annu. Rev. Econ. 4, no. 1: 131-157.
Hanushek, Eric A., John F. Kain, and Steven G. Rivkin. 1999. “Do higher salaries buy better
teachers?” NBER Working Paper 7082
Hedges. Larry V., Richard D. Laine, and Rob Greenwald. 1994. “An Exchange: Part I: Does
Money Matter? A Meta-analysis of Studies of the Effects of Differential School Inputs on
Student Outcomes.” Educational Researcher, 23(5): 5–11.
Hoxby, Caroline M. 2000. The Effects of Class Size on Student Achievement: New
Evidencefrom Population Variation. The Quarterly Journal of Economics, 115(4): 1239–
85.
Jackson, C. Kirabo, Jonah E. Rockoff, and Douglas O. Staiger. 2014. "Teacher Effects and
Teacher-Related Policies." Annu. Rev. Econ. 6, no. 1: 801-825.
Kane, Thomas J., Jonah E. Rockoff, and Douglas O. Staiger. 2008. "What does certification tell
us about teacher effectiveness? Evidence from New York City." Economics of Education
Review 27, no. 6: 615-631.
Krueger, Alan B. 1999. “Experimental Estimates of Education Production Functions.”
Quarterly Journal of Economics, 114(2): 497–532.
Krueger, Alan B. 2003. “Economic Considerations and Class Size.” The Economic Journal,
113(485): F34–F63.
Krueger, Alan B., and Diane M. Whitmore. 2001. “The Effect of Attending a Small Class in the
Early Grades on College-Test Taking and Middle School Test Results: Evidence from
Project Star.” The Economic Journal 111 (468): 1–28.
MacPhail-Wilcox, Bettye, and Richard A. King. 1986. "Production functions revisited in the
context of educational reform." Journal of Education Finance, 191-222.
Murnane, Richard J. 1975. The Impact of School Resources on the Learning of Inner
City Children. Cambridge, MA: Ballinger.
Murnane, Richard J., Judith D. Singer, John B. Willett, James Kemple, and Randall Olsen.
1991. “Who will teach? Policies that matter.” Cambridge: Havard University Press.
Rivkin, Steven G., Eric A. Hanushek, and John Kain. 2005. “Teachers, Schools and
Academic Achievement.” Econometrica, 73(2): 417–58.
Rockoff, Jonah E. 2004. “The Impact of Individual Teachers on Student Achievement:

29
 Evidence from Panel Data.” American Economic Review, 94(2): 247–52.
Rockoff, Jonah E. 2009."Field experiments in class size from the early twentieth century." The
Journal of Economic Perspectives: 211-230.
Summers, Anita A., and Barbara L. Wolfe. 1977. "Do schools make a difference?" American
Economic Review, 67(4), 639-652
Word, Elizabeth, John Johnston, Helen P. Bain, B. DeWayne Fulton, Jayne B. Zaharias,
Martha N. Lintz, Charles M. Achilles, John Folger, and Carolyn Breda. 1990.
“Student/Teacher Achievement Ratio (STAR): Tennessee’s K–3 Class Size Study, Final
Report.” Nashville, TN: Tennessee State Department of Education. Quarterly Journal of
Economics, 114(2): 497–532.

30
 X. Appendix

TABLE 1 – Variable Description

Variable Description Duration

Student-teacher Ratio for school Annually data from school year 2004 to
STRi,t
district i at year t school year 2013

School enrollment for school district Annually data from school year 2004 to
TotalStudentsINDistricti,t
i at year t school year 2013

Total Amount of budget for school Annually data from school year 1999 to
TotalDistYearBgti,t
district i at year t school year 2013

% of unknown race for school Annually data from school year 2000 to
UnknownRacei,t
district i at year t school year 2014

% change of Pacific Islands

Annually data from school year 2000 to
PacificIslanderi,t Americans for school district i at
school year 2014
year t

% of African Americans for school Annually data from school year 2000 to
Blacki,t
district i at year t school year 2014

% of Asian Americans for school Annually data from school year 2000 to
Asiani,t
district i at year t school year 2014

% of Native Americans for school Annually data from school year 2000 to
NativeAmericani,t
district i at year t school year 2014

% of Hispanic Americans for school Annually data from school year 2000 to
Hispanici,t
district i at year t school year 2014

Amount of teachers in school district Annually data from school year 2000 to
TeacherINDistricti,t
i at year t school year 2014

Average years of teaching

Annually data from school year 2000 to
yrsindistcnti,t experience for school district i at
school year 2014
year t

% of students receiving free and

Annually data from school year 2000 to
distfreeredlnchpcti,t reduced lunch for school district i at
school year 2014
year t

31
 Table 2 – Final Ordinary Least Squares Regression Results

(1) (2) (3) (4) (5)

VARIABLES Log (The Log (Athletics) Log (PE) Log (Extra) Log (STR)
Arts)

Log (Years in District) -0.276* -0.271*** -0.404** -0.431** 0.102***

(0.141) (0.0687) (0.198) (0.177) (0.0370)
Log (District Yearly Budget) 0.124 0.296** 0.524*** 0.214 -0.138***
(0.0916) (0.126) (0.154) (0.209) (0.0301)
Log (Students In District) 0.113 0.0688 0.519 0.250 0.861***
(0.149) (0.0696) (0.429) (0.278) (0.0585)
Unknown Race 0.00704 0.111 0.323 -0.104 -0.00609
(0.164) (0.0711) (0.208) (0.200) (0.0368)
Pacific Islander -2.186 4.151 1.522 9.825 -0.0140
(2.817) (3.953) (6.774) (8.781) (0.621)
Black 1.325 3.940 4.726 -0.581 0.490
(2.237) (4.056) (5.888) (6.829) (0.536)
Asian 0.725 -3.060** -6.193* 8.014* -0.571
(1.379) (1.191) (3.335) (4.468) (0.424)
Native American 1.223 -1.640* -2.262* -2.091 0.121*
(0.940) (0.880) (1.167) (1.588) (0.0690)
Hispanic 0.629 -0.0465 -0.550 0.637 -0.0383
(0.582) (0.223) (0.544) (0.430) (0.0960)
District Free Lunch % -0.00140 0.00175 -0.00279 0.00779** -0.000190
(0.00360) (0.00261) (0.00405) (0.00354) (0.000549)
Constant 10.90*** 7.910*** 0.637 5.808** -1.578***
(2.141) (2.264) (4.380) (2.785) (0.554)

Observations 1,525 1,226 1,403 1,021 1,740

R-squared 0.923 0.896 0.884 0.853 0.932
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1

32
FIGURE 5:

The percentage change as a response to a 10% increase in

Average Teacher Experience
2.0%

1.0%
1.0%

0.0%

-1.0%

-2.0%

-3.0% -2.7% -2.8%

-4.0%
-4.0%
-4.3%
-5.0%
Athletics Physical Extra Curricular The Arts STR
Education

Hypothesis & Methodology Equations:

Continued from Hypothesis & Methodology section, the specific equations used in our

empirical analysis are as follows.

4.1.1. The effect of Teacher Experience on the Budget Allocation for Physical Education:

33
4.1.2. The effect of Teacher Experience on the Budget Allocation for Athletics:

4.1.3. The effect of Teacher Experience on Budget Allocation for the Arts:

4.1.4. The effect of Teacher Experience on the Budget Allocation for Extracurricular

Activities:

34
4.2.1. The effect of Teacher Experience on Student-Teacher Ratios:

35