Abstract
This paper examines the relation between teacher pay and teacher quality through the career dynamics of teachers and non-teachers. I find that public school teachers earn considerably less than their comparable college graduates in the non-teaching sector. By tracking wage differentials before and after career changes, I find evidence of positive selection, in which high-paid teachers are more likely to move to non-teaching occupations, and of negative selection, in which low-paid non-teachers tend to move to the teaching sector. These selection patterns, which ultimately contribute to a decrease in teacher quality, are more significant in union-unfriendly states.
There exist ongoing debates over whether teachers are underpaid or not. Some assert that public school teachers are “overpaid,” considering the decrease in teacher quality and stagnant academic achievement of students. Biggs and Richwine (2011) claim that there is no teacher wage penalty when they compare teachers and non-teachers with similar Armed Forces Qualification Test (AFQT) scores. However, the National Education Association (NEA) research division contends that teachers have been “underpaid” compared with other college graduates in non-teaching industries, and that, in 2000, the wage gap between non-teaching careers and teaching careers among college graduates was about 60 percent for males and 16 percent for females (Hurley 2004). During recent teacher strikes in Arizona, Colorado, Oklahoma, Virginia, and West Virginia, teachers asked for a pay increase claiming that they are so little paid that they must work multiple jobs to make ends meet. These seemingly opposite views toward the wage gap between teachers and non-teachers motivate this study.
In 2018, the Supreme Court ruled in Janus v. American Federation of State, County and Municipal Employees (Janus hereafter) that the agency fees paid by nonunion members in the public sector violated First Amendment rights protecting freedom of speech and association. This dramatic legal change, immediately impacting unions’ financial capacity, may have influenced teacher compensation and shifted the career decisions of individuals who otherwise may have chosen a teaching profession. Therefore, understanding what unions have done for the wage gap and the career dynamics between the teaching and the non-teaching sectors under different legal environments toward teachers unions during the pre-Janus period could help policy makers better grasp the role of teachers unions in the post-Janus era.
This study investigates the relation between teacher pay and teacher quality through the career dynamics of teachers and non-teachers. The main goals of this paper are threefold. First, I assess and establish the extent to which teachers’ earnings are lower or higher than other comparable non-teachers’ earnings. Second, I analyze the career paths of teachers and non-teachers by using changes in earnings for job switchers to infer differences in worker quality between the two occupational groups. Finally, I examine the role of teachers unions in the process of career changes by comparing outcomes among different legal environments toward unions.
The primary data source for this study is the Current Population Survey (CPS) Merged Outgoing Rotation Group between 2001 and 2018. I also use the CPS Supplements for Occupation Mobility and Job Tenure, which has detailed information on holders of multiple jobs and occupation mobility. As a complement to the CPS, I exploit the School and Staffing Survey (SASS) for 2007-2008 and 2011-2012 and their Teacher Follow-Up Survey (TFS), administered by the National Center for Education Statistics (NCES).
I use a two-sector version of the Roy Model to study the changes in earnings associated with the career dynamics of teachers and non-teachers. The model predicts positive selection, in terms of worker quality such that the teachers who move into the non-teaching sector are more likely to have higher ability, and negative selection, in which non-teachers who move into the teaching sector tend to have lower ability. Greater teacher compensation in pro-union states may provide an incentive for high-quality teachers to remain in the teaching sector while encouraging some high-quality non-teachers to join the teaching sector. Thus, the model ultimately predicts that both positive and negative selection of career-changers will be more noticeable in less unionized states than in highly unionized states.
I estimate the wage gap between teachers and non-teachers using the state and time fixed effects models. Then, I employ the individual fixed effects with two time periods (the first difference model) to compare differences in earnings followed by career changes of teachers and non-teachers. Finally, to extrapolate worker quality, I compare earnings before career changes within the legal environments toward teachers unions.
Literature
Disappointed with their students’ performance in international tests, educators and school officials in the United States have been endeavoring to improve the quality of public education. In particular, the quality of teachers has been pointed out as the key to successful educational reform. If teachers are truly underpaid, and if the best teachers have attractive options outside teaching, improving teacher quality seems almost unachievable with the limited resources.
Fortunately, several recent studies find that high-quality teachers respond to incentives. Biasi (2019) finds that districts that adopted flexible pay are more likely to have teachers whose value-added is higher because young, high value-added teachers who were underpaid relative to their productivity left the rigid-pay districts and joined flexible-pay ones that paid according to one’s effectiveness. Macartney, McMillan, and Petronijevic (2018), using rich longitudinal data from North Carolina, show that teachers improve their value-added under incentive-oriented policies. Based on a discrete-choice experiment, Johnston (2018) finds that teachers with high value-added prefer merit pay more than those with low value-added. Given these findings that economic incentives play an important role in attracting and retaining high-quality teachers, comprehending teachers’ decisions on career paths will shed light on how to structure teacher compensation that can raise teacher quality and improve educational outcomes.
Numerous studies have examined the relationship between teacher pay and educational outcomes, but empirical evidence produced has been mixed. Earlier studies often found that teacher salaries have either insignificant or negligent effects on student outcomes (Betts 1995; Hanushek 1986). Hanushek (2003) shows that, out of the 118 estimates, 20 percent showed a positive coefficient, 7 percent negative, and 73 percent were statistically insignificant. More recently, Hanushek (2015), who uses meta-analysis, claims that there exists weak support for the claim that simply raising teacher pay or overall educational spending will improve educational outcomes.
However, more recent evidence that is based on alternative methods and natural experiments tends to show that higher teacher pay leads to positive student outcomes. Loeb and Page (2000) estimate that a 10 percent increase in teacher pay reduces dropout rates by 3 to 6 percent. A study by RAND Education (2006), using a natural experiment based on the types of districts, shows that higher teacher salaries were associated with higher student performance in Illinois. Hendricks (2014), using data from Texas between 1996 and 2012, finds that a 1 percent increase in teacher pay reduces teacher turnover by 0.16 percentage points and also shows that paying teachers more improves student achievement through higher retention rates. Britton and Propper (2016), using a natural experiment in England, find that teachers respond to pay, and a 10 percent shock to the wage gap between the local labor market and teacher pay leads to an average loss of 2 percent in average school performance. When using the cross-country variations of teacher compensation among Organisation for Economic Co-operation and Development (OECD) countries, Dolton and Marcenaro-Gutiérrez (2011) estimate that a 15 percent increase in teacher salaries raises students’ test scores in Trends in International Mathematics and Science Study (TIMSS) by 6 to 8 percent. Numerous studies on teacher effectiveness find a significantly positive association between teacher quality and teacher salary (Ballou and Podgursky 2001; Figlio and Lucas 2004; Flyer and Rosen 1997; Goldin 2006; Murnane et al. 1991; Temin 2002).
One of the hypotheses that may explain the inconsistent findings regarding the effect of teacher pay on educational outcomes is that the variation in teacher salaries may be too little to induce any significant change in teacher quality, due to a limited geographic scope of data. Studies using state-level data are more likely to find significant effects of teacher salaries on teacher quality than those using district-level data within the same state, because within-state variation tends to be smaller than cross-state variation in teacher pay. 1 To address this issue, therefore, my study looks at all fifty U.S. states to examine the relation between teacher pay and teacher quality.
Several studies directly compute the wage gap between teachers and non-teachers to examine if teachers earn less than other comparable professions (Allegretto and Mishel 2016; Hurley 2004; Murnane et al. 1991; Temin 2002), but most of them report the wage differentials without considering other wage determinants beyond the basic individual characteristics.
Many studies find that teachers respond to incentives. Biasi (2019) finds that districts that adopted flexible pay ended up with teachers whose value-added was higher. Part of the improvement arose because young, high value-added teachers, who were systemically underpaid given their productivity, left the rigid-pay districts and joined flexible-pay ones where they could be paid more in proportion to their contributions.
Researchers find that average teacher quality has decreased over time in the United States. Murnane et al. (1991) and Bacolod (2003) show a drop in the fraction of college graduates with high AFQT scores among teachers. Using U.S. Census data, Lakdawalla (2001) shows that the relative schooling of teachers declined by about three years from the 1900 birth cohort to the 1950 birth cohort. Corcoran, Evans, and Schwab (2004) find that the average quality of female teachers, measured by the quartile ranking based on their placement in the distribution of mathematics and verbal exams administered in their senior year of high school, has fallen since 1960. Hoxby and Leigh (2004) find that female teachers’ academic aptitude in the United States has decreased. They point out that increasing outside options, which pay higher salaries to highly productive females who once entered the teaching profession, have drawn them to non-teaching careers.
More recently, Goldhaber and Walch (2014) spot a recent gain in teacher quality, measured by average Scholastic Assessment Test (SAT) percentile rankings for teachers for 2008 college graduates entering the teacher workforce the following school year. Similarly, Nagler, Piopiunik, and West (2017) find that teachers entering the profession during recessions are significantly more effective in raising student test scores. Ingersoll, Merrill, and Stuckey (2014) find that the quality of female teachers, if assessed by college selectivity, has been constant (not falling) for the past two decades, although there has been a continuous decrease in the proportion of male teachers from top institutions. These new findings suggest that teacher quality among new entrants may vary, depending on specific periods or general business cycles. However, as long as public schools continue to face a higher attrition rate among novice teachers, and if new teachers with higher aptitude are more likely to leave teaching, it may contribute to a decrease in the average teacher quality.
Building upon the literature on teacher pay and teacher quality, my study aims to answer three specific questions. First, are teachers facing a wage penalty? Second, how do teachers and non-teachers with different abilities make career decisions, and who are these career-shifters? Third, what is the impact of legal environments toward teachers unions on career dynamics and teacher quality?
The Roy Model for Career Changes
If teachers are paid less than comparable non-teachers, they should be able to attain a larger increase in their earnings when they move to non-teaching occupations than non-teachers who change to other non-teaching occupations. The fundamental problem to this approach is that the career change is an endogenous outcome of various optimizing decisions, in which people self-select the occupational sector that provides them with the highest expected earnings. To deal with the self-selection issue of career-changers, I use the Roy Model that is inspired by Borjas (1987), who used simple two-sector settings to discuss the self-selection of immigrants. Supplemental Appendix I provides the full description of the model.
Consider two occupational groups: teachers (A) and non-teachers (B). Let
Suppose that there is a cost in changing careers, C, and it is proportional to
Teachers will move into the non-teaching sector if
Descriptive Statistics by Occupational Group and Gender.
Source. CPS MORG, 2001-2018.
Standard deviation in parentheses. Only college graduates who are working full-time (usual work hours with thirty-five and above), ages eighteen to sixty-four, are included in the sample. Non-teachers group excludes private school teachers. CPS MORG = Current Population Survey Merged Outgoing Rotation Group.
Negative selection occurs if non-teachers are negatively selected in the ability distribution of non-teachers, and they are also below the mean of the ability distribution of teachers when they move into the teaching sector. The two conditions necessary for this negative selection to happen are the following: (1) the non-teaching sector (origin) has a more dispersed wage distribution than the teaching sector (destination), and (2) the correlation between the ability that is relevant both in the teaching and non-teaching sectors is also sufficiently high. As these conditions are met as described above, the Roy Model predicts that non-teachers with low ability will want to move into the teaching sector to take advantage of the narrower wage distribution.
The higher teacher pay that unions negotiate reduces the teacher wage penalty, providing an incentive for high-quality teachers to remain in the teaching profession while encouraging some high-quality non-teachers to join the teaching force. Thus, I predict that both positive and negative selection of career-changers will be weak in highly unionized states. In contrast, states with weak unions will find many of their high-quality teachers leaving classrooms for non-teaching occupations and very few high-quality non-teachers joining the teaching sector. These selection patterns, therefore, will be more noticeable in less unionized states.
Data and Method
My primary data source is the Current Population Survey Merged Outgoing Rotation Group (CPS MORG) files for 2001 through 2018. The CPS MORG is not fully panel data, as households are not tracked once they move. Every household that enters the CPS is interviewed each month for four months, then ignored for eight months, and then interviewed again for four more months. Thus, we observe the same individuals in the same household twice yearly if they do not move. I pool monthly data into the annual data for eighteen consecutive years.
I restrict the sample to full-time workers who worked last week with the usual work hours per week of thirty-five or above, are ages eighteen to sixty-four, and possess college degrees or above. Public school teachers are defined as primary school, secondary school, and special education schoolteachers. I focus on people with at least college degrees, because most public school teachers have bachelor’s degrees. To make a clear comparison between public school teachers who remain in teaching and those who completely leave the teaching sector, the non-teachers group excludes private school teachers. The non-teachers group includes school principals and other administrators. 5 Supplemental Appendix II provides the summary statistics from the CPS MORG for 2000-2018.
Following Allegretto and Mishel (2016, 2018), my empirical analysis of teachers’ relative earnings focuses on comparisons of weekly earnings, instead of annual salaries, to avoid measurement issues that arise because teachers’ contracts are typically nine months. I also look at hourly earnings, the weekly earnings on the primary job divided by usual hours worked per week on the job. The earnings data are truncated at the bottom and top 1 percentile to eliminate large outliers.
I also use the data from the CPS Supplements for Occupation Mobility and Job Tenure (SOMJT), which provides additional information about holders of multiple jobs, such as the number of jobs held, characteristics of the second jobs, and occupational mobility. 6
As a complement to the CPS, I use the 2007-2008 and 2011-2012 Schools and Staffing Survey (SASS) as well as the 2008-2009 and 2013-2014 TFS. The SASS is administered by the National Center for Education Statistics (NCES), and it is large-scale and nationally representative data. About a third of U.S. public school districts are included in the SASS. One year after the SASS is collected, teachers who left the teaching sector are surveyed, forming the TFS for Former Teachers.
Table 1 provides descriptive statistics for public school teachers and non-teachers from the CPS MORG. On average, both weekly and hourly earnings are lower for public school teachers than for non-teachers. The fraction of female workers and earnings ratio of females to males are much higher for public school teachers. A greater proportion of public school teachers hold master’s degrees, but a lower proportion hold professional school or doctorate degrees. Compared with public school teachers, a larger fraction of non-teachers lives in metropolitan regions. A majority of public school teachers are unionized, but only 10 percent of non-teachers join a union. More than 95 percent of public school teachers hold U.S. citizenship, and the proportion is 10 percentage points lower for non-teachers.
Until the recent efforts to restrict the collective bargaining (CB) rights of public school teachers in some states and the Janus decision, the legal environments for teachers unions, which reflect states’ general perspective toward unions, have been stable for the past several decades. For instance, states that mandated CB four decades ago still do so in 2018.
Following Moe (2011), I define the legal environment toward teachers unions using two legal criteria, outlined in pre-Janus settings: whether a state allows CB of public school teachers and whether it permits agency fees so that nonunion teachers pay their “fair share” fees for the services unions provide (which is no longer applicable after Janus). Based on these two legal frameworks, I classify the fifty U.S. states into four groups.
The first group, which I call the High-CB group, is composed of twenty-one states that have duty-to-bargain laws and allow unions and school districts to negotiate mandatory agency fees for nonunion members. About half of public school teachers belong to this group. The second group, the Med-CB group, also mandates CB but prohibits agency fees. The third group, the Low-CB group, permits but does not require districts to sign CB contracts. The fourth group, the No-CB group, bans CB of public school teachers.
Since 2011, Idaho, Indiana, Kentucky, Tennessee, Michigan, and Wisconsin have substantially restricted teachers’ CB rights, and they belong to different groups after their change in state legislation. 7 To deal with this complication, I drop these states and examine if the teacher wage penalty varies, depending on the legal environment. Figure 1 presents the map of the group categorization between 2001 and 2018 to visualize how these states are spread across the United States.
Legal environment toward teachers unions.
Empirical Strategies
The first part of my empirical analysis estimates the wage gap between non-teaching and teaching careers, controlling for various wage determinants. This approach can provide direct evidence if teachers are paid less, relative to their comparable college graduate peers in the non-teaching sector. I use the following model to estimate the wage differential:
where i, s, and t indicate teachers, states, and time periods, respectively. Teacher is a dummy variable for public school teachers,
The second part of my empirical studies focuses on the changes in earnings over a one-year period and investigates the relationship between earnings and career paths in the teaching and non-teaching sectors. The Roy Model predicts that positive selection occurs when teachers move into the non-teaching sector, but negative selection occurs when non-teachers move into the teaching sector. Highly productive teachers are more likely to move into the non-teaching sector to take advantage of higher returns to ability whereas non-teachers with low ability are more likely to move into the teaching sector with a narrower earnings distribution.
If public school teachers are truly underpaid relative to comparable non-teachers, they must have better alternative opportunities in the non-teaching sector and obtain large increases in their earnings after the career change. To verify this, I assess the changes in earnings associated with the career dynamics of teachers and non-teachers.
During a one-year period, some people, whom I call “career-changers,” change their careers while others, called “career-stayers,” remain in their occupations. The changes in earnings of career-stayers can be interpreted as returns to experience of current occupations. The key assumption in this approach is that most wage determinants do not change or change little in one year. For instance, career changes from the teaching sector to non-teaching professions in the fields of finance or law will require additional investment in human capital and will not easily occur within a one-year time period. The only thing that can considerably change is their career, which is the driving force for their wage changes. I compare the differences in earnings during the one-year period by using the individual fixed effects model with two time periods, which is equivalent to a first difference model:
where
It is noteworthy that teachers usually face a uniform salary schedule, which mainly depends on two observable credentials: experience and education. Thus, it is difficult to assess teacher productivity/quality based on teachers’ base salary schedule alone, because it is only associated with teacher quality to the extent those credentials are associated with true quality. Furthermore, literature has found no significant relationship between teacher quality and education level, only a modest positive relationship with experience, and mixed results with teacher certification status (Clotfelter, Ladd, and Vigdor 2010; Harris and Sass 2011; Kane, Rockoff, and Staiger 2008).
Nevertheless, earnings can still serve as a proxy for worker ability because they can measure teachers’ capability in obtaining higher pay in public schools, and the earning capability of teachers within the teaching sector is arguably correlated with their earning capability outside the teaching sector (Chingos and West 2012). In fact, as in the Roy model, my study assumes that teachers’ earning ability, not their productivity/effectiveness per se, is the key factor that influences their decisions on career change. Thus, even if pay “level” is a weak proxy for teacher productivity, it is the “change” in earnings after a career change that is more relevant in inferring teacher quality.
School districts reward teachers based on qualifications that are designed to improve teacher quality, such as National Board for Professional Teaching Standards certificates 9 and Highly Qualified Teacher (HQT) 10 status. According to the 2007-2008 and 2001-2012 SASS, more than 40 percent of districts reward teachers who have attained National Board Certificates and HQT by providing incentives such as cash bonuses and salary increases. Approximately 10 percent of school districts reward excellence in teaching, and more than 5 percent of public school teachers report that they earn any additional compensation from their school system based on their students’ performance through a merit pay or pay-for-performance agreement.
The earnings of non-teachers reflect their abilities, as the non-teaching sector usually employs an incentive pay system. Therefore, I use the total weekly earnings, which is the sum of base salaries and additional compensations that are closely related to worker performance. The total weekly earnings can better assess worker quality beyond what base salaries can measure. 11
As mentioned before, Chingos and West (2012) show that the correlation between the ability that is relevant in both the teaching and the non-teaching sectors is high so that teachers with high ability will be more likely to be more productive workers in the non-teaching sector. Therefore, comparing the earnings before and after career changes within and across occupational groups can capture the difference in worker quality. Most of the difference in ∆ across occupations can be attributed to the variance in worker productivity if we compare among workers with similar experience and education level. The CPS does not provide data about how long one has been working at a current job, so it is difficult to know exactly how much work experience one has accumulated. Moreover, the costs of transitioning careers are likely to vary with tenure in the chosen profession. The attrition rate is high, mostly driven by poor initial matching quality, among younger cohorts in any labor market, and the costs of switching careers tend to be lower, as their job-specific human capital investment is relatively low. To deal with these issues, I examine ∆ by age group in which workers are more likely to share a similar level of experience. 12 I assess ∆ for four career paths for various age groups:
Stay in the same teaching position
Change from a public school teacher to a non-teaching occupation
Stay in the same non-teaching occupation
Change from a non-teaching occupation to a public school teacher
where two paths (one and three) describe career-stayers and the other two career-changers (two and four).
Examining changes in earnings after career changes helps identify who changes their career. However, whether teachers who move into the non-teaching sector are high-ability people and whether non-teachers who move into the teaching sector are low-ability people remains uncertain. Thus, I compare workers’ earnings reported during the first interview, which I call “pre-earnings,” to indirectly assess the quality of workers. If teachers who move to the non-teaching sector had significantly higher pre-earnings than teachers who stay in their careers, this suggests that the former have relatively higher ability than the latter. If non-teachers who move into the teaching sector had substantially lower pre-earnings than non-teachers who stay in their occupations, it suggests that the former have relatively lower ability than the latter.
It should be pointed out that this study assumes that the primary reason for which low-skilled workers move into the teaching sector or high-skilled teachers move into the non-teaching sector is economic incentives. This may not be the case for all career shifters. Thus, this study may not provide a full assessment for career dynamics when nonpecuniary factors motivate the career changes.
To examine the role of teachers unions in the career decision and earning dynamics of teachers, I re-estimate models (5) without state dummies and (8) by different legal environment toward teachers unions and compare outcomes to make inferences on worker quality.
Results
The Teacher Wage Penalty and Wage Determinants
Table 2 reports the results by age group and gender. Overall, public school teachers earn 6.5 percent less compared with other college graduates who have similar wage determinants but work in the non-teaching sector. 13 For the all ages group, the teacher wage penalty is 9.5 percent for male teachers and 5.7 percent for female teachers. The wage penalty is smaller for teachers under thirty years old, but it becomes substantial for teachers in their thirties and remains high in their forties. Teachers in their fifties and above face the smallest wage penalty. 14
Wage Penalty of Public School Teachers by Age Group and Gender. Dependent Variable: Log(Weekly Earnings).
Source. CPS MORG, 2001-2018.
Errors are clustered within states (presented in parentheses). Only college graduates who are working full-time (usual work hours with thirty-five and above), ages eighteen to sixty-four, are included in the sample. Non-teachers group excludes private school teachers. Control variables include a dummy for master’s degree, a dummy for professional or doctorate degree, potential experience and its square, usual hours of work per week, categorical dummies for race, ethnicity, marital status, primary family relationship, union status, citizenship status, metropolitan area, and population size. All regressions employ state fixed effects and year fixed effects. All regressions use persons’ composite final weight. CPS MORG = Current Population Survey Merged Outgoing Rotation Group.
p < .1. **p < .05. ***p < .01.
As a robustness test, I drop potentially endogenous wage determinants, such as family head status, union membership, citizenship status, metropolitan status, and population size of the region from the regressions. The alternative results are generally similar to those presented in Table 2. However, the coefficients after dropping those variables are somewhat greater (more negative) for male teachers, although the new coefficients are very similar to those in Table 2 for female teachers. This suggests that the potentially bad control variables effectively erode some of the estimated teacher wage penalty, especially for male teachers (see Table A in Supplemental Appendix III for summarized results). One may point out that the absence of fringe benefits in the analysis leads to an overestimation of the teacher wage penalty. Considering that new entrants into the labor market often place more weight on salary than on benefits when choosing an occupation, the wage penalty estimated using weekly/hourly earnings provides a good assessment, at least, for young people with limited work experience. 15 In addition, a study by Fitzpatrick (2015) finds that a majority of employees in Illinois Public Schools who value their pension benefits are willing to trade only twenty cents for a dollar’s worth of future benefits, providing evidence that the actual compensating wage differential between salaries and fringe benefits that teachers prefer may be much smaller than what the public perceives. For a comprehensive analysis, however, I discuss the implication of non-wage benefits in the frame of compensating wage differentials in the Discussion section.
The teacher wage penalty is likely to force teachers to work multiple jobs. Figure 2 presents the percent of people with multiple jobs in the teaching and non-teaching sectors by gender, based on the author’s calculation using the SOMJT. Because teachers may take on additional jobs during the summer, I exclude summer months (June through August). As expected, a greater fraction of teachers holds multiple jobs than non-teachers. Approximately 16 percent of male teachers are multiple jobholders, but only 5 percent of male non-teachers have more than one job. Among females, 9 percent of teachers and 5 percent of non-teachers report multiple jobs. 16
Multiple jobholder, by occupational group and gender.
To see which factors contribute to the teacher wage penalty, I investigate the relationship between wage determinants and earnings by occupational and age groups. Table 3 reports the estimated coefficients and standard errors of major wage determinants. Panel A presents the results for public school teachers and Panel B for non-teachers.
Wage Determinants by Occupational and Age Group. Dependent Variable: Dependent Variable: Log(Weekly Earnings).
Source. CPS MORG, 2001-2018.
Errors are clustered within states (presented in parentheses). Only college graduates who are working full-time (usual work hours with thirty-five and above), ages eighteen to sixty-four, are included in the sample. Non-teachers group excludes private school teachers. Other control variables include potential experience2, usual hours of work per week, categorical dummies for marital status, primary family relationship, citizenship status, and population size. CPS MORG = Current Population Survey Merged Outgoing Rotation Group; FE = fixed effects.
p < .1. **p < .05. ***p < .01.
The returns to higher education (master’s, professional school, or doctorate degree) are greater for non-teachers than for teachers. For people under thirty, returns to master’s degree are about 14 percent for teachers and 19 percent for non-teachers. For the same age group, the returns to professional school or doctorate degree are 21 percent for teachers and 30 percent for non-teachers. This gap in returns to education may be partly due to the different types of advanced degrees obtained by teachers and non-teachers (e.g., the EdD for teachers vs. MBA, JD, MD, and PhD for non-teachers). Due to the lack of detailed information on education degrees, I am unable to state how much of this differential is attributed to the teacher wage penalty.
Compared with the non-teaching sector, the teaching sector has lower wage inequality across gender, race, and ethnicity. Male workers have significantly higher weekly earnings than female workers in both occupational sectors (see Table 1), but the gender gap in the non-teaching sector is twice as large as that in the teaching sector. Female teachers earn more comparable wages to their male counterparts than female non-teachers do. On average, in the non-teaching sector, the non-white, except young Asians, has significantly lower earnings than the white. In the teaching sector, black teachers earn significantly less than white teachers, but this gap is much smaller than the gap in the non-teaching sector. Hispanics have substantially lower earnings than the non-Hispanics in the non-teaching sector but not in the teaching sector.
It is commonly believed that public school teachers work fewer hours per week so that they earn larger hourly earnings than non-teachers. However, my data show that public school teachers earn much less for each hour they put in per week than their comparable non-teachers do. The additional hour per week is associated with higher weekly earnings by 0.44 percent for public school teachers and by 1.33 percent for non-teachers. This is not linked to the fact that teachers get summers off because I find the same results when I drop the summer months from my observations. The reason why the returns to working hours per week are significantly lower for teachers may be because teachers do not receive overtime pay. 17
The returns to experience is lower for teachers than for non-teachers. For age under thirty, one year of experience is associated with a 2.3 percent increase in weekly earnings for teachers and 5.1 percent for non-teachers. This evidences why the teacher attrition rate is so high among new teachers (Carroll and Foster 2010).
The nominal income should be higher in metropolitan areas to compensate for higher living costs and to keep the real income stable. However, teachers receive a smaller premium (9.3%) than non-teachers (15.6%) who live in the same metropolitan regions. Thus, teachers in urban areas face an even greater wage penalty than teachers in rural areas, and this will aggravate the problem of low teacher retention in urban regions, which face difficulty recruiting teachers.
There is a substantial difference in the average union membership rate between public school teachers and non-teachers (see Table 1, 64% vs. 9%). Not surprisingly, the union wage premium is substantial only in the teaching sector.
One may be concerned that placing principals or administrators in the non-teachers group may artificially inflate the changes in income after the career change from teachers to non-teachers. As a sensitivity test, I drop school principals and administrators from the sample and redo the analysis. The patterns of alternative results remain the same as presented in Tables 2 and 3.
For an additional sensitivity analysis, I also use the log of hourly earnings instead of weekly earnings. The alternative results are very similar to those presented in Tables 2 and 3. The magnitude and the statistical significance of the wage penalty are slightly greater for female teachers (see Tables B and C in Supplemental Appendix III for summarized results).
Career Changes between the Teaching and Non-teaching Sectors
The teacher wage penalty and teachers’ higher rates of multiple jobholding may serve as a push factor for teachers and alter average worker qualities via career changes between the teaching and non-teaching sectors. Table 4 presents the summary of changes in earnings during the one-year period, estimated from the individual fixed effects model (4) for four career paths by age group. The first panel presents the results for people under thirty, and the second, third, and fourth panels for people in their thirties, forties, and fifties and above, respectively. In each panel, the first two rows present career dynamics and changes in earnings for public school teachers, and the next two rows are for non-teachers. To control for the hours of work per week and to deal with higher earnings of non-teachers working overtime, I focus on changes in hourly earnings. The change in earnings for those who stay in their careers establishes the baseline for the analysis because it represents the returns to experience of one’s current career for a one-year period.
The Changes in Earnings over a Period of One Year, by Age Group.
∆ = Log(hourly earnings)2 – Log(hourly earnings)1.
Source. CPS MORG, 2001-2018.
Only college graduates who are working full-time (usual work hours with thirty-five and above), ages eighteen to sixty-four, are included in the sample. Non-teachers group excludes private school teachers. Standard errors are reported in parentheses. CPS MORG = Current Population Survey Merged Outgoing Rotation Group.
I first discuss career dynamics and changes in earnings for public school teachers. The first panel of Table 4 shows changes in earnings for people in their twenties. The returns to experience of teachers (1.3%) is smaller than those of non-teachers (8.4%). The low returns to experience for young teachers contributes to the low teacher retention rate among novice teachers. Public school teachers who move into the non-teaching sector in their twenties experience a big raise in their earnings (12.4%) as they find far better earning opportunities outside the teaching sector. Panel 2 shows that teachers in their thirties who move into the non-teaching sector also obtain 8.5% increases in their hourly earnings. It is noteworthy that these estimates in increase in earnings understate the full advantage of leaving the teaching sector because the non-teaching sector offers higher returns to experience, and I concentrate only on the changes in earnings after one year.
Teachers in their forties and above have higher returns to experience than non-teachers of the same age group. For instance, hourly earnings rise by 3.6 percent (4.2%) for teachers but by 2.9 percent (2.5%) for non-teachers in the forties (fifties) age group. This shows that teacher pay, if measured in hourly earnings, is back-loaded, which puts great weight on seniority. Thus, older teachers are better off staying in the teaching sector.
Non-teachers who move into the teaching sector in their twenties and thirties receive higher earnings by 8.3 percent and 4.4 percent, respectively. Considering that teachers in these age groups face a wage penalty (see Table 2), non-teachers who still decide to become teachers are more likely to be low-ability people with less successful career tracks in the non-teaching sector. Non-teachers in the older age groups are also better off staying in their current occupations.
I also examine the changes in earnings by potential experience group. The new results are qualitatively similar to those reported in Table 4. For people with less experience (twenty years and less), teachers gain a large increase in their earnings when they leave the teaching sector. Non-teachers with twenty years and less experience gain a considerable increase in their earnings when they move into the teaching sector. The returns to experience for the greater experience group (between twenty and above) is larger for public school teachers than for non-teachers. 18
What Unions Do for the Career Dynamics of Teachers and Non-teachers
To investigate the relation between the teacher wage penalty and teachers unions, I reestimate the wage gap between teachers and non-teachers by different legal environment. Table 5 presents the results by legal group and gender. In the High-CB group, only male teachers face a small wage penalty. Female teachers earn less than female non-teachers, but the difference is statistically insignificantly different from zero. Public school teachers, both male and female, in other legal environments face a substantial wage penalty, ranging between 8 percent and 16 percent. Table D in Supplemental Appendix III reports alternative results based on hourly earnings.
Wage Penalty of Public School Teachers by Legal Environment and Gender. Dependent Variable: Log(Weekly Earnings).
Source. CPS MORG, 2001-2018.
Errors are clustered within states (presented in parentheses). Only college graduates who are working full-time (usual work hours with thirty-five and above), ages eighteen to sixty-four, are included in the sample. Non-teachers group excludes private school teachers. The High-CB group is composed of twenty states that have duty-to-bargain laws and allow unions and school districts to negotiate mandatory agency fees for nonunion members. The Med-CB group also has duty-to-bargain laws but prohibits agency fees. The Low-CB group allows local school districts to sign CB agreements but does not require them to bargain with unions. The No-CB group bans public school teachers from collectively bargaining. Idaho, Indiana, Kentucky, Tennessee, Michigan, and Wisconsin are excluded due to the change in their state legislation toward teachers unions during the study period. Control variables include a dummy for master’s degree, a dummy for professional or doctorate degree, potential experience and its square, usual hours of work per week, categorical dummies for race, ethnicity, marital status, a dummy for family head, union status, citizenship status, metropolitan area, and population size, and year dummies. All regressions use persons’ composite final weight. CPS MORG = Current Population Survey Merged Outgoing Rotation Group; CB = collective bargaining.
p < .1. **p < .05. ***p < .01.
Table 6 summarizes the earnings dynamics of career-changers for workers under the age of 40 by legal environment. I focus on hourly earnings to control for working hours per week. Again, the High-CB group shows a very different pattern from the other three groups. In the High-CB group, teachers who leave for the non-teaching sector earn 5.5 percent higher hourly earnings, and non-teachers who become public school teachers increase their earning by more than 10 percent. Because teachers do not face a significant wage penalty (see Table 5), high-quality teachers may be less likely to leave the profession, and some high-quality non-teachers may move into the teaching sector.
The Changes in Earnings over a Period of One Year, by Legal Environment. Younger Group (Age <40 Years Old)
∆ = Log(hourly earnings)2 – Log(hourly earnings)1.
Source. CPS MORG, 2001-2018.
Only college graduates who are working full-time (usual work hours with thirty-five and above), ages eighteen to sixty-four, are included in the sample. Non-teachers group excludes private school teachers. The High-CB group is composed of twenty states that have duty-to-bargain laws and allow unions and school districts to negotiate mandatory agency fees for nonunion members. The Med-CB group also has duty-to-bargain laws but prohibits agency fees. The Low-CB group allows local school districts to sign CB agreements but does not require them to bargain with unions. The No-CB group bans public school teachers from collectively bargaining. Idaho, Indiana, Kentucky, Tennessee, Michigan, and Wisconsin are excluded due to the change in their state legislation toward teachers unions during the study period. Standard errors are reported in parentheses. CPS MORG = Current Population Survey Merged Outgoing Rotation Group; CB = collective bargaining.
In other groups where teachers face a large wage penalty, however, it is teachers moving to the non-teaching sector who see a substantial gain (11%-18%) in their earnings. Non-teachers obtain a smaller increase (even negative in the Low-CB group) in their earnings after moving into the teaching sector. Due to the large wage penalty, high-quality teachers may be more likely to leave their careers and find better opportunities in the non-teaching sector. High-quality non-teachers may find little incentive to join the teaching force.
Both the High-CB and Med-CB groups mandate CB, but the Med-CB group shows more similarity with other groups that do not have compulsory bargaining laws. This is consistent with Han (2019), who shows that the union effects on teacher outcomes in the states that do not allow agency fees is much smaller than those in the states allowing agency fees. She attributes the low union impacts in the states with no-agency fees to the higher incidence of free riding of nonunion teachers who are covered by the same union contract, because free riding is likely to cause substantial damage to unions’ financial strength.
Selection among Career-Changers and Worker Quality
The Roy Model predicts that high-quality teachers are more likely to move into the non-teaching sector (positive selection), and that non-teachers with low ability are more likely to move into the teaching sector (negative selection). Tables 4 through 6 show that young teachers are better off moving into the non-teaching sector, and some young non-teachers decide to join the teaching force even when typical teachers face the wage penalty. However, it is not yet clear whether the teachers moving into the non-teaching sector are people with high ability and whether the non-teachers moving into the teaching sector are people with low ability.
In Table 7, I compare the earnings people reported during the first interview, which I call “pre-earnings,” which offer supporting evidence for this selection pattern predicted by the model. Table 3 has revealed that individuals’ education, work experience, and metropolitan regions are the three key factors that make the different impact on the wages of young teachers and non-teachers. Thus, I compare the conditional pre-earnings by focusing on workers under the age of forty who have master’s degrees and live in metropolitan regions.
Conditional Pre-earnings, by Occupational Group and Legal Environment for Younger Group (Age <40 Years Old).
Source. CPS MORG, 2001-2018.
Only college graduates who are working full-time (usual work hours with thirty-five and above), ages eighteen to sixty-four, are included in the sample. Non-teachers group excludes private school teachers. The High-CB group is composed of twenty states that have duty-to-bargain laws and allow unions and school districts to negotiate mandatory agency fees for nonunion members. The Med-CB group also has duty-to-bargain laws but prohibits agency fees. The Low-CB group allows local school districts to sign CB agreements but does not require them to bargain with unions. The No-CB group bans public school teachers from collectively bargaining. Idaho, Indiana, Kentucky, Tennessee, Michigan, and Wisconsin are excluded due to the change in their state legislation toward teachers unions during the study period. Pre-earnings is the earnings that people reported during the first interview. Standard errors are reported in parentheses. CPS MORG = Current Population Survey Merged Outgoing Rotation Group; CB = collective bargaining.
Panel A presents the average pre-earnings of public school teachers by career path and legal environment. Teachers who move into the non-teaching sector are, on average, more likely to have higher earnings than teachers who stay in their teaching professions, which supports the positive selection. In all groups, public school teachers who leave the teaching sector report higher pre-earnings than those who stay in their careers by $1.57, which is statistically significant at the 5 percent level. As expected, this positive selection is nonexistent in the High-CB group but more apparent in the other three groups.
Panel B summarizes the pre-earnings for non-teachers under forty. Overall, non-teachers who become public school teachers have significantly lower pre-earnings ($25.47) than non-teachers who remain in their careers ($28.95), and the difference is significant at the 1 percent level. This provides evidence for negative selection; young non-teachers with lower earning ability are more likely to move into the teaching sector. Many of these young non-teachers with master’s degrees living in metropolitan areas experience a large increase in their earnings when they switch to the teaching sector. The negative selection is more conspicuous in the other three groups than in the High-CB group. 19
To have a better understanding of the full story of career dynamics and changes in earnings, I look at the earnings distribution for teachers and non-teachers using the box plot. The left and right sides of the box represent the 25th percentile and 75th percentile of the distribution, respectively, and the vertical line near the middle of the box indicates the 50th percentile, the median. Figure 3A displays the box plots of pre-earnings and after-earnings 20 for teachers under forty by career path. Because the box plots are heavy-tailed and skewed to the right, it is vital to look at the median of the distribution. The distribution of pre-earnings of teachers who stay in their careers (the first box plot) is slightly to the left of the distribution of pre-earnings of teachers who move into the non-teaching sector (the third box plot). The after-earnings distribution for teachers who leave teaching (the fourth box plot) is further to the right than the after-earnings distribution of career-staying teachers (the second box plot). The medians of both pre-earnings and after-earnings are greater for teachers who move into the non-teaching sector than those of teachers who stay in their careers.
Earnings distribution before and after career change.
Figure 3B shows the box plot for young non-teachers. The pre-earnings distribution of non-teachers who move into the teaching sector (the third box plot) is considerably to the left of the pre-earnings distribution of non-teachers who stay in their careers (the first box plot). The after-earnings distribution of non-teachers who become teachers (the fourth box plot) is also to the left of the after-earnings distribution of non-teachers who stay in their careers (the second box plot). The differences in medians between the two career paths are quite substantial.
To examine the role of unions in positive selection, Figure 4A presents the pre-earnings distribution for teachers by career path for each legal environment. The pre-earnings distribution for young public school teachers who move into the non-teaching sector (second box plot for each group) is to the right of the pre-earnings distribution of teachers who stay in their careers (first box plot for each group), and the median is also higher for teachers leaving the teaching sector, with the exception of the Med-CB group. In the Low-CB and No-CB groups, there exists a significant positive selection, encouraging high-quality teachers to leave the teaching sector.
Earnings distribution before and after career change, by legal environment.
In Figure 4B, in all groups, non-teachers who become public school teachers have lower median pre-earnings, and the distribution is noticeably to the left of that for non-teachers who stay in their careers. The gap in the median pre-earnings between the two career paths is much smaller for the High-CB group than for other groups, indicating that the negative selection occurs at a lesser degree in the High-CB group.
Then, who are these teachers and non-teachers changing their occupational groups? According to the CPS SOMJT, the top ten career picks for public school teachers who display the positive selection pattern when moving to the non-teaching sector include managers (in various branches), education administrators, scientists or engineers, postsecondary teachers, lawyers (and other judicial workers), accountants, sales representatives, preschool or kindergarten teachers, registered nurses, and counselors. Many of the non-teachers who have held these occupations also move into the teaching sector. Surprisingly, among new teachers, some used to work at low-skilled jobs such as cashiers, janitors or building cleaners, dining room helpers, cargo and freight agents, and movers. For these people, teacher wages may appear attractive, although teachers actually face a wage penalty. This might suggest that the teacher hiring process is too lenient, and that it bolsters the mechanism of negative selection.
Many teachers remain in the education sector even after they change occupations. As a sensitivity analysis, I drop principals and other educational administrators from the non-teacher group and redo the analysis. The alternative results under this new specification are similar to those already presented. The patterns regarding teacher wage penalty remain qualitatively and quantitively the same. When teachers move to the non-teaching sector, the increase in earnings becomes somewhat larger in all four groups. The pre-earnings of non-teachers who have master’s degrees and remain in their non-teaching occupations become higher in all groups. These findings suggest that even principals may have been underpaid, compared with non-teachers, in our public school system. 21
Discussion and Conclusion
This paper examines the relation between teacher pay and teacher quality through career changes. I examine the career paths of teachers and non-teachers and changes in earnings for a one-year period to conjecture differences in worker quality. I then analyze the role of teachers unions by comparing the career dynamics and the selection pattern by different legal environment toward teachers unions.
This study finds that public school teachers are paid less, relative to comparable college graduates working in the non-teaching sector, partly because they face lower returns to education and experience compared with non-teachers. The teacher wage penalty is linked to a higher prevalence of multiple jobholding among teachers than among non-teachers.
Applying the Roy Model, this research shows that high-paid public school teachers tend to leave the teaching profession in the early stages of their careers, as they find it more beneficial to move into the non-teaching sector, which is the case of positive selection. In contrast, low-paid non-teachers are more likely to move into the teaching sector, which shows the pattern of negative selection.
According to the theory of compensating wage differentials, teachers may earn lower salaries than non-teachers, in exchange for better fringe benefits. If what we observe is a simple case of compensating differentials, every public school teacher should have the same chance to make greater hourly earnings in exchange for less fringe benefits when they move into the non-teaching sector, and there is no reason to find the significant difference in pre-earnings between teachers who remain in teaching and those who leave teaching, especially after controlling for key wage determinants. By the same argument, we should not observe a significant difference in the pre-earnings between non-teachers who stay in the non-teaching sector and those who move into the teaching sector. I find a significant difference, however, in the pre-earnings between career-stayers and career-changers in each occupational group. This result strongly suggests that career-changers and career-stayers belong to different parts of the ability distribution of the relevant population in each occupational group, and compensating differentials alone cannot fully explain the wage gap between teachers and non-teachers.
Compensating differentials also imply that teachers should earn more salary and fewer benefits when they move to the non-teaching sector, such that the total compensation for teachers and non-teachers who share the same wage determinants are commensurate. If this is the case, the smaller wages teachers receive should not be considered a “penalty.” If, however, teachers face a true wage penalty, their total compensation (salary plus benefits) would be lower than that of non-teachers with the same ability. In this case, when teachers move to the non-teaching sector, they will be able to receive a large increase in total compensation. Fringe benefit is just one employment condition, and there will be other features in employment that concern compensating differentials in a broad definition. Job security, for instance, is often cited as a privilege of teaching professions, as some teachers may be willing to work for less money in exchange for higher job security.
According to the 2008-2009 and 2012-2013 TFS for Former Teachers, approximately 8 percent of teachers left the teaching sector after the base survey year, and 7 percent of those leavers identified that compensation was the most important factor for their departure. About 50 percent of teachers who left for the non-teaching sector indicated that their current non-wage benefits and job security are similar to what they received in the teaching sector. Approximately 25 percent of those teachers even reported that non-teaching occupations offered better non-wage benefits and job security than teaching careers. 22 These findings reject the common belief that teachers have better fringe benefits and job security, so that they are willing to accept lower salaries. At least 75 percent of teachers who move to non-teaching occupations claim that the fringe benefits of their new jobs are at least as substantial as their former teaching jobs. Thus, the survey suggests that the wage penalty that teachers face may be even greater if the amount of teachers’ non-wage benefits is incorporated into the analysis.
The teacher wage penalty may create a vicious cycle in the educational system, which leads to a detrimental equilibrium of teacher salaries and teacher quality. Low teacher salaries will not be able to attract bright young minds into the teaching sector, so the teacher quality will start low as the average quality of teacher applicants’ pool will be low. The returns to experience for these novice teachers is smaller than those for non-teachers, so teachers with high productivity are very likely to leave teaching before they accumulate ten years of experience. Through the mechanism of positive and negative selection in career dynamics, teachers with low productivity are more likely to remain in the teaching sector while non-teachers with low ability are more likely to move into the teaching sector. Therefore, average teacher quality falls even more (or remains low if the exit and entry rates of high-quality teachers are the same), and the low teacher quality justifies low salaries. Consequently, the education system gets stuck at an inferior equilibrium where a low level of salaries is matched with low teacher productivity.
The teacher wage penalty leads to a teacher shortage. While the student population is predicted to increase by about three million students in the next decade, teacher education enrollment dropped by 35 percent between 2009 and 2014 (Sutcher et al., 2016). Strauss (2017) reports that every U.S. state is facing an intensified shortage of teachers in key subject areas, and some states, such as Oklahoma, Utah, and Arizona, have started to hire teachers without formal training to fill the gap. Darling-Hammond et al. (2016) find that, in California, where teacher demand is rising, the number of teachers hired on substandard permits and credentials doubled, comprising a third of all new credentials issued in 2014-2015. My study also finds a clear pattern of negative selection where non-teachers with low pre-earning ability are more likely to move into the teaching sector. Over time, therefore, the teacher shortage problem is likely to further reduce teacher quality. The teacher shortage problem substantially varies by region and teacher type, and it is most acute for science, technology, engineering, and mathematics (STEM) subjects, possibly due to the greater availability of STEM-related opportunities outside of teaching (Boe and Cook 2006; Goldhaber et al. 2015). This suggests that the wage penalty will be even greater for STEM teachers than for non-STEM teachers.
States that are challenged by teacher recruitment are attempting to convince former teachers to return to the teaching profession. According to the author’s calculation based on the 2008-2009 and 2012-2013 TFS, more than 80 percent of teachers who left the teaching sector reported that an increase in their salary would be important (somewhat, 21 percent; very, 26 percent; and extremely important, 34 percent) in influencing their decision to return to the position of a K-12 teacher. This provides additional evidence that raising teacher pay is crucial for reducing the teacher shortage.
When facing teacher shortage problems, districts may lower their standards for teacher qualifications (for instance, alternative certification is available via online teacher preparation programs, which can be completed in as little as one year in many states). If, however, non-teaching careers have been requiring higher skills and productivities than teaching careers, teachers who move out of the teaching sector would have to be high in their ability distribution, providing strong context for positive selection. Non-teachers who move into the teaching sector, instead, will more likely be from the low portion of the ability distribution of the non-teaching sector, which is the pattern of negative selection.
Another key finding of this study is that both positive and negative selection occur predominantly in the states that are hostile to unions. This suggests that average teacher quality will fall in the anti-union states, but less likely so in the pro-union states whose teachers do not face a significant wage penalty. The High-CB group offers greater incentives for high-quality teachers to remain in the teaching sector, even though many of those teachers still move into the non-teaching sector for better earning opportunities. Some high-quality non-teachers may be willing to move into the teaching sector in the High-CB group, whereas few in the other three groups will consider becoming teachers. As a result, the average teacher quality will be higher in the High-CB group than in the other groups. This is echoed in a recent study by Han and Maloney (2019) who find that students in unionized districts have significantly higher standardized test scores in both math and English.
As income inequality has been rising in the recent period, and a union presence tends to raise average teacher pay while compressing the variance in their earnings, this study predicts that both positive and negative selection problems will become exacerbated in the anti-union states, because it will become easier for high-quality teachers to find better opportunities in the non-teaching sector whereas the teaching profession will look more attractive to low-quality non-teachers who favor smaller variances in earnings.
Moreover, this study also finds that the Med-CB group, which mandates CB but bans agency fees, shows similar patterns with the Low-CB and No-CB groups, suggesting that agency fees collection is an important institution for teachers unions. In the post-Janus period, therefore, I expect that teachers in the Med-CB group will encounter what the Low-CB and No-CB teachers have been experiencing, and quality will fall as the teacher wage penalty rises due to the limited bargaining power of teachers unions.
The policy implication of my study is twofold: to improve teacher quality, (1) raise total teacher compensation significantly enough to attract people with high productivity into the teaching sector, and (2) increase the returns to experience for young teachers.
Any education policy that attempts to raise teacher quality with the current pool of teacher candidates may not bring a significant and steady improvement. Teacher compensation must increase and the pay raise must be substantial enough to change the composition of the current applicants’ pool. The results in Table 4 suggest that teacher pay must increase at least about 11 percent (12.4 minus 1.3) for teachers under the age of thirty and about 4 percent (8.5 minus 4.4) for teachers in their thirties, to encourage them to stay in their current teaching positions. This is aligned with the proposal by Temin (2002) who also argues that the United States is currently stuck at the lower level of equilibrium and cannot escape from it unless it pays considerably more to teachers. Leigh (2012) finds that higher teacher pay does encourage more able people to enter a teaching career: a 1 percent increase in the salary of a starting teacher raises the average aptitude of students who are entering teacher education programs by 0.6 percentile ranks.
The seniority-based salary schedule should be modified to increase the returns to experience for young teachers. In the educational sector, returns to experience is not linear. The literature finds that the marginal returns to teaching experience diminish such that teacher effectiveness increases more quickly during the first several years of teaching, then wanes over time (Boyd et al., 2007; Ladd 2008; Rockoff 2004). Thus, wage structures that put more weight on young teachers might create more marginal benefits in the educational system. A recent study by Grissom and Strunk (2012) finds that school districts with teacher compensation schemes that allocate greater resources to novice teachers than to veteran teachers, called “front-leading,” display better student performance.
This study only focuses on the labor income of workers due to the data limitation of the CPS. Future research should consider workers’ total compensation, including non-wage benefits, and other employment issues, such as working conditions and job security, to fully understand the relationship between career dynamics across occupational sectors and worker quality. The lack of information on the college path, majors of teachers and non-teachers, and types of teachers (primary vs. secondary) forces this study to assume that all bachelor’s degrees are equivalent, preventing me from conducting a more elaborate analysis. This topic on the association between college path and career changes is also left for future study.
Supplemental Material
Online_Appendix_R2 – Supplemental material for Teacher Wage Penalty and Decrease in Teacher Quality: Evidence from Career-Changers
Supplemental material, Online_Appendix_R2 for Teacher Wage Penalty and Decrease in Teacher Quality: Evidence from Career-Changers by Eunice S. Han in Labor Studies Journal
Footnotes
Acknowledgements
I thank Elaine Bernard, Raj Chetty, Richard Freeman, Jeffrey Keefe, Jerry Marschke, Marty West, seminar participants at Harvard University and colleagues at the University of Utah for their many helpful comments. I thank the National Bureau of Economic Research (NBER) for providing me with the necessary facilities and assistance. I also thank the National Center for Educational Statistics (NCES) for kindly providing me with the data. The views expressed herein are my own and do not necessarily reflect the views of the NBER or the NCES.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.
Notes
Author Biography
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
