Teacher Pensions Blog

  • I recently read a story about the statewide teacher salary schedule in North Carolina. Politicians there are debating various changes for the upcoming year, but what caught my eye was the weird way North Carolina pays its teachers. 

    I pulled the graphic below, from a blog by The Progressive Pulse, to show what I mean. The graph shows the current salary schedule (in red) versus two competing proposals from the Governor and state Senate. My focus was on the red line. North Carolina teachers earn very small raises (of 0.5 to 2.5 percent) from years 1-14, and then they earn a 4.6 percent raise upon reaching 15 years of experience, then nothing until year 20, when they get a 6 percent bump, and then nothing again until year 25, when they get another a 6.25 percent bump. I added the arrows pointing to these weird kinks. 

    It's easy to imagine how North Carolina got to this place, but those back-end lumps are not tackling North Carolina's real turnover problems. Like every other state, the bulk of North Carolina's teacher turnover happens in the early years, and the vast majority of its incoming teachers will never reach these back-end bonuses (the state estimates only 28 percent will reach the 15-year mark). Worse, these back-end salary bumps are also not doing much to shape the retention decisions for those teachers who do reach them, at least not according to the state's pension plan. 

    To show what I mean, I pulled the graphic below from the Actuarial Valuation report from the Teachers’ and State Employees’ Retirement System of North Carolina. I drew a red circle over North Carolina's official estimates for how many teachers will leave each year, based on their age. North Carolina has a separate set of assumptions for teachers in their first four years on the job, but it switches to these age-based estimates after that. (The chart can be read as, "For 25-year-old females, North Carolina assumes 11 percent will leave the state's public schools in that given year.)

    Note that, while North Carolina's statewide salary schedule is based on years of experience, the state pension plan hinges the bulk of its assumptions on age, not years of experience. That is, after four years, the pension plan does not feel the need to account for the salary schedule's age-based longevity incentives. Even if there were a connection to years of experience, the state is not assuming any bumps in retention rates tied to any particular age.

    To put it bluntly, North Carolina's pension plan does not assume that its statewide teacher salary schedule is affecting turnover rates. 

    This isn't unusual. The New York City teacher salary schedule has weird kinks at years 10, 13, 15, 18, 20, and 22. For example, after receiving no raise in year 21, a New York City teacher earns a $5,511 raise in year 22, a bump of 5.8 percent. But New York City's own pension plan doesn't assume this matters enough to affect its financial situation. Here are the corresponding withdrawal assumptions for the Teachers’ Retirement System of the City of New York: 

    This is a pretty straightforward chart. (It can be read, "New York City assumes 9 percent of teachers with 0-5 years of service will leave within the next year.") New York City is assuming that teacher turnover rates fall every five years. This is a much less incremental approach than the assumptions North Carolina uses, which makes it even more obvious that New York City's salary bumps at years 10, 13, 15, 18, 20, and 22 are not doing enough to shape teacher behavior to warrant adjusting the pension plan's assumptions. 

    Now, it's possible that teachers are responding to these lumpy incentives, and the pension assumptions simply aren't fine-grained enough to detect them. But that means something too. Those pension plan assumptions are the basis for consequential financial decisions about how much the state or city needs to save today in order to pay benefits in the future. While states and cities are spending a lot of money on back-end incentives like this, their own pension plans don't think it's worth altering their assumptions to account for them. 

  • My colleague Kelly Robson and I have a new paper out in Education Next this week looking at the interaction between pension plans and teacher retention. We had two main findings. One, pension plans themselves do not assume that teachers change their behavior in order to qualify for a pension. And two, while there may be some late-career retention effect as teachers at the end of their career hold on in order to maximize their pension, state pension plans assume a much larger "push-out" effects that causes large numbers of veteran teachers to retire at relatively young ages. 

    I'll come back to the second point in a subsequent post, but I'll start with the lack of early-career retention effects. If qualifying for a pension were an incentive, we should see teachers change their behavior in order to "vest" and reach their state's minimum threshold. That is, teacher turnover rates should flatten out in the years leading up to the vesting period, and then spike upwards, at least somewhat, immediately afterwards. For example, if teachers in a given state qualified for a pension after five years, teachers in their fourth year should be marginally less likely to leave their jobs as they near that point. After all, these teachers would qualify for a guaranteed stream of pension income every month upon retirement if they stay just one more year. Then, some teachers who held on to year five solely to qualify for a pension would leave the profession and retention rates would rise.

    We tested this empirically using the pension plans' own financial assumptions. Each state pension plan publishes “withdrawal” rate tables estimating the percentage of teachers who will leave (aka withdraw from) the pension system in a given year. State pension plans publish these withdrawal rate assumptions in their Comprehensive Annual Financial Reports (CAFRs). Though these rates reflect each state’s predictions for the future, states update them regularly based on their own historical patterns.

    States present their withdrawal assumptions in tables similar to the figure below, which is an example from Alabama. The table can be interpreted as follows: For 25-year-old female teachers with 0 years of service, Alabama assumes that 14 percent will leave by the end of their first year. The state assumes another 14 percent will withdraw when this cohort of teachers is 26, 27, 28, and 29 years old. When they reach age 30 with five years of teaching experience, the state assumes the annual withdrawal rate for this group drops to 5.8 percent. 

    This table can be used to determine whether Alabama believes its 10-year vesting requirement to qualify for a pension is shaping teacher behavior. If it were, we should see slightly lower turnover rates in years eight and nine, as teachers hold out for a pension, followed by a small spike in year 10 as teachers who were remaining solely to qualify for a pension finally departed. But that’s not what Alabama assumes. Across all ages and experience levels, Alabama expects withdrawal rates (for non-retirement purposes) to decline the longer a teacher remains in the profession. It does not assume there will be holdouts prior to the pension threshold who leave upon qualifying for it. Again, Alabama is making these assumptions based on its own historical data.

    In fact, after looking at every state's assumptions, we could find no evidence of any state pension plan that believes its teachers will systematically change their behavior to qualify for a pension. Instead, teachers become less and less likely to leave the profession (for non-retirement reasons) every year they remain, and there is no indication that qualifying for a pension affects that trend one way or another. Given the large numbers of teachers who leave early in their career, about 60 percent of Alamaba teachers and about half of all new teachers nationwide don't stick around long enough to qualify for a pension. Although they and their employer are contributing large sums of money to the pension plan, they won't earn a pension at all, and the pension plan won't be enough to keep them in the profession. That's bad for those teachers in terms of retirement savings, and it's bad for employers who could have used that money in more productive ways.

    In a subsequent post I'll turn back to later-career teachers. In brief, pension plans do appear to exert a limited “pull” effect that keeps some late-career teachers on the job (remember, most teachers have left before then). But pension plans exert an even stronger “push” effect that encourages veteran teachers to retire, regardless of any teacher’s particular interest or ability to continue teaching. Because veteran teachers tend to perform better than a replacement teacher just entering the profession, that has real, negative consequences for students. 

  • At the end of April a Cook County judge dismissed a lawsuit brought by Chicago Public Schools against Illinois Governor Bruce Rauner. CPS alleged that both the state school funding formula and the teacher pensions system are unconstitutional because they lead to the systematic underfunding of the education of low-income students and students of color.

    While Judge Franklin Valderrama recognized that the state’s school finance system is broken, he nevertheless dismissed the case. Plaintiffs do have the opportunity, however, to refile by May 26.

    This puts CPS right back where it was when the suit was filed: facing a massive budget deficit that may force the district to end the school year weeks early. Unable to pay its bills, the state legislature passed a bill late last year allocating over $215 million to help CPS keep schools open for the full year.

    The problem is that Governor Rauner vetoed the bill. He refused to send the necessary emergency funds to Chicago to alleviate a problem – one the state helped to create – without first winning significant reforms to the teacher pension system. In effect, Governor Rauner is holding Chicago students hostage. He is leveraging roughly 3-weeks of education for hundreds of thousands of students in exchange for pension reforms.

    To be clear, the state’s finances are an unmitigated disaster and in desperate need of reform. But, the ends do not justify the means. Coercing the legislature to reform pensions by threatening harm to Chicago’s students smacks of extortion. But the gambit may be paying off for Gov. Rauner. The legislature is currently considering a bill to reform teacher pensions that may appease the governor and encourage him to release the millions for CPS.

    There is a worry that the Speaker of the House, Michael Madigan, who himself hails from Chicago, will derail the process. Some view Madigan as a longstanding obstacle to important reforms in CPS. That view may be right in this instance. But the strategy of using students as a bargaining chip likely strengthened Madigan’s position.

    Consider the recent example from Pennsylvania, the former Governor of Pennsylvania, Tom Corbett unsuccessfully employed a similar approach by withholding roughly $50 million from Philadelphia School District without concessions from the Philadelphia Federation of Teachers. Governor Rauner may find a similar outcome here: threatening students is a poor political strategy.

    The financial woes facing Chicago Public Schools are considerable. Fixing them requires compromise on both sides. But bargaining with students’ education is a recipe for disaster and may make reform even more difficult.   

  • Giving teachers a say over their retirement plan just might be good for everyone.

    The majority of states enroll their public school teachers in defined benefit (DB) pension plans. These plans are back-loaded, and they mainly benefit the small portion of teachers who remain both in the classroom and in the same state for 20 years or more. Supporters of these plans argue pensions are a retention tool – teachers might be less likely to leave the profession if there’s a large financial incentive waiting for them if they stay. These advocates rarely acknowledge that this idea suggests it's ok to use someone's retirement security as a tool to shape their behavior, but it's worth investigating whether these claims play out in pratice. That is, does a teacher's retirement plan shape her behavior?

    My colleague Chad Aldeman has a forthcoming piece on this question in Education Next, but a recent study published by the National Center for Analysis of Longitudinal Data in Education Research (CALDER) provides some initial answers. The study, authored by Dan Goldhaber, Cyrus Grout, and Kristian Holden, looked at what happened when Washington State switched from a pure DB system to a hybrid plan. Hybrid plans combine aspects from portable, defined contribution (DC) models, like 401(k)s, as well as traditional, defined benefit plans, like most pensions. Here’s how it worked:

    Washington introduced their current hybrid plan (called TRS3) in 1995, to replace their existing DB plan (TRS2). At the time, existing employees remained in the original DB plan, while new hires were enrolled in the new hybrid plan. A two-year transfer period allowed teachers in the DB plan the option to switch into the hybrid plan – those who did so received a bonus payment equal to 65% of their accrued TRS2 contributions.

    In 2007, Washington made changes again. Teachers hired after that point can select either the DB plan or the hybrid plan, with the latter as the default option for those who do not make an affirmative selection. The table below shows the differences between the two plans: 

    Public Pension Reform and Teacher Turnover: Evidence from Washington State 2015

    So what happened to teacher retention after Washington changed its retirement plan? The study compares three sets of teacher turnover rates:

    1.    Teachers hired just before and after the introduction of the hybrid plan

    2.    Teachers who could choose between the DB or hybrid plan as new hires

    3.    Teachers who could transfer from the DB plan to the hybrid plan

    In all three of these situations, proponents of pensions as retirement incentives would expect higher turnover rates from those teachers enrolled in TRS3, the hybrid plan. In this plan, overall compensation is less back-loaded, which decreases the monetary incentive for teachers to stick it out for a big payout at retirement.

    In the first two comparisons, there was no systemic difference in turnover patterns between the two groups. The last comparison though, examining those teachers who transferred from the DB pension plan to the DC hybrid plan does show a relationship between pension structure and teacher turnover. But instead of turnover increasing with a portable plan, it actually decreased. From 1998-2005, teachers who transferred into the hybrid plan actually had turnover rates that were 1-4 percentage points lower than those who remained in the DB plan. The authors hypothesized that perhaps giving teachers a choice made teaching more attractive.

    Overall, this study presents additional evidence that teachers simply aren’t that sensitive to pension plan design. Further, in another paper from Goldhaber and Grout, they found that Washington’s hybrid plan did not harm teacher quality or retirement security. In practice, the pensions-as-retention-strategy talking point doesn't hold up. 

  • Federal data from the National Center on Education Statistics (NCES) offers a potentially surprising revelation: Private school teachers have higher turnover rates than their public school counterparts, and it’s not particularly close.

    The data below capture what NCES calls the “leaver” rate. NCES regularly surveys teachers, and it divides respondents into three categories: stayers, movers, and leavers. Stayers are teachers who were teaching in the same school in the current school year as in the base year. Movers are teachers who were still teaching in the current school year, but who had moved to a different school. And the leavers, represented in the graph, are teachers who left the profession entirely.

    As the graph shows, the teacher leaver rate is almost twice as high at private schools than it is at public schools. Both have increased over time, but private schools have seen their rates increase even faster. These data call into question many of the common explanations for changes to teacher turnover rates among public school teachers, such as No Child Left Behind, teacher evaluation reforms, or the Common Core. Those reforms, which applied primarily to public schools, simply can't explain the increases in teacher turnover in private schools. (In fact, during the NCLB era, public school teacher turnover did rise a bit, but private school turnover rose even more.)

    Source: NCES private/public

    The next graph shows how the "leaver" rates have changed over time for public and private school teachers, by their years of experience. Over time, private schools have seen dramatic increases in turnover among early-career teachers, whereas in the public sector, early-career teachers are more likely to stay today than they were in the late 1980s. In fact, private school leaver rates have accelerated faster than public school rates for every age group except those with 20 or more years of experience. (As my colleague Chad Aldeman has written, those late-career turnover increases can be traced at least partially to changing demographics and rising retirement rates.) 

    Source: NCES

    Another way to look at this data is to attempt to follow synthetic "cohorts" of teachers over time. We've run this same analysis on public school teachers and found that cumulative retention rates for public school teachers haven't changed that much over time. Regardless of the year they started, about one-third of public school teachers had left within five years, and about half were gone within 10 years.

    But compare that finding to private school teachers, where we see a noticable difference across cohorts. Rather than the lines overlapping, signaling similar turnover rates, we see clear gaps across entry years. In the private sector, unlike in public schools, teachers who entered in 1987 had higher retention rates than teachers who were hired in 1990, and so on. Those gaps are smaller in more recent years, but the NCES data suggest that far fewer private school teachers today are making it to key career milestones than did in the past. 

    Since the cohort graph is somewhat hard to read, see the same data in the table below. Each column represents a starting year, and the rows indicate the cumulative retention rate by years of experience. Private school teachers who leave within three years of experience provide an especially compelling example. Among new private school teachers in 1987, a little over one-quarter had left within three years. In 2008, more than half were gone within the same time frame. Again, these are much higher figures than for public school teachers. 


    Since this is a pension blog, it's worth mentioning that we do NOT think pensions are the cause of, or solution to, this issue. First, while public sector teachers are more likely to be enrolled in defined benefit pension plans, that disparity existed in the 1980s as well. That is, it can't explain the changes over time, nor can it explain the changes by age group. Second, while pensions could theoretically boost teacher retention, in practice we don't actually see much evidence of that. We don't have plausible theories for why turnover in private schools seems to be rising much faster than it is in public schools--although we'd love to hear suggestions--but as the country considers making additional public investments in private schools, it's worth wondering why these schools are losing so many of their teachers. 

  • Florida offers its teachers a choice. When they begin working in Florida schools, they can choose to join the state's traditional defined benefit (DB) pension plan, or they can enroll in a portable defined contribution (DC) plan instead. The state has an entire website devoted to helping teachers decide which plan is best for them given their age, how long they plan to stay, and how comfortable they are investing money.

    At first glance, Florida seems neutral about which option teachers choose. When I took the quiz to identify which plan would be better for me, it recommended the portable DC plan and reassured me that there were a range of investing options, even for people who weren't that confident in their investing abilities. Another state document has a nifty chart estimating which teachers would be better off in which plan, depending on their starting age and how long they planned to stay. It looks like this: 

    The state's own estimates suggest that anyone who starts teaching in Florida under the age of 45 would be better off in the portable "Investment Plan." Even for people who begin teaching later in life, they would benefit from the DB Pension Plan only if they stayed more than 8 but less than 23 years. So, Florida is essentially telling teachers that there's only a small sliver of teachers for whom the DB Pension Plan would be a better option. 

    Another document on the Florida Retirement System site includes a footnote that specifically warns teachers that, "According to FRS historical statistics, less than 20% of newly hired employees and 50% of those with over 10 years of service actually stay a full career in FRS employment, given today's mobile society." 

    Indeed, this lines up with my own research on Florida's DB plan. Only about one-quarter of new Florida teachers remain in the pension plan for the eight years it takes to qualify for any pension plan at all. A Florida teacher must teach continuously for 24 years before finally qualifying for a pension worth more than her own contributions plus interest. And only about one-in-ten stick around long enough to reach the state's normal retirement age. In short, Florida's defined benefit pension system doesn't work well for the vast majority of its teachers. 

    Why, then, does Florida default all of its teachers into the defined benefit Pension Plan? (Note: See update below. Florida has since changed its default option.) If teachers do not proactively enroll in a retirement plan within their first five months on the job--a time when many first-year teachers are more worried about the demands of their new job--the state automatically enrolls them in the Pension Plan. To put it another way, Florida defaults all of its rookie teachers into a retirement plan that, as the plan itself acknowledges, is probably not right for most of them. 

    Now, Florida deserves credit for offering its teachers a choice at all. Alaska is the only state that automatically enrolls all teachers in a portable retirement plan, but five other states in addition to Florida--Michigan, Ohio, South Carolina, Utah, Washington--provide teachers a choice.  The rest automatically place all of their teachers into back-loaded defined benefit plans. 

    But Florida is unlikely to see substantial enrollment changes until it shifts its default. Defaults can be powerful "nudges" that encourage people into behaviors they may not otherwise proactively choose. When companies have switched their retirement plans to automatic participation (with optional opt-outs), they have seen enormous enrollment gains, even though the choices remain the same. These patterns also apply to things like contribution rates and investment choices.

    Florida should be applying those lessons to help nudge teachers into better decisions. A bill currently making its way through the state legislature would do just that. It would still give Florida teachers a choice over their retirement plan, but it would set the more portable option as the default. Rookie teachers may not have the time or the wherewhithal to think about their retirement plan, but the state should nonetheless help them make smart decisions. 

    In the summer of 2017, Florida passed legislation to shift the default option to the portable defined contribution plan, following the recommendations in this post. The new rule applies to all Florida teachers hired after January 1, 2018. For more information, see here