Human Capital Policies in Education: Further Research on Teachers and Principals 5rd Annual CALDER Conference January 27th, 2012
Pension-Induced Rigidities in the Labor Market for School Leaders Cory Koedel Jason A. Grissom Shawn Ni Michael Podgursky
Motivation • Most educators participate in defined-benefit (DB) pension plans that penalize mobility. • These penalties will introduce rigidities into the labor market that coincide with pension borders (usually state lines). • The extent to which pension borders actually affect mobility is an empirical question. • Why school leaders? • A burgeoning research literature points to school leadership as a key determinant of student success (Branch et al., 2011; Brewer, 1993; Cannon et al., 2010; Clark et al., 2009; Coelli and Green, 2010; Dhuey and Smith, 2010; Grissom and Loeb, 2011). • Leadership work typically occurs when mobility penalties in pension systems are the largest.
Key Results • We first document the costs of crossing a pension border for a representative new school leader in Missouri (coming from teaching). • The costs are in the hundreds of thousands of dollars as measured by the present discounted value of stream of pension payments. • We then identify the effect that these costs have on interschool mobility near pension boundaries. • For two groups of schools that are separated by a pension border, our most conservative estimates indicate that removing the border will increase leadership flows by approximately 100 percent. • This is a big deal in Missouri, the location of our study, because the two urban districts in the state operate their own pension systems. • The findings will generalize to other pension-border regions (essentially state lines). • Finally, we consider the policy consequences of the restricted mobility in Missouri. • the pension borders separate the city school districts from a more-qualified leadership pool on the outside. • Causality (for leadership quality) is unclear. But, even if the pension borders do not cause the gap in leadership quality that we observe in the data, it is indisputable that they represent a major impediment to eliminating it.
Pension Borders and Leadership Quality • Border Effects • Decrease size of applicant pools in border regions • Exacerbate or maintain inequities in leadership quality on different sides of a border • Reduce the quality of leader-school matches (Jackson, 2010 – teacher context)
Why Missouri? • Missouri is unique in that educators belong to three distinct pension systems: PSRS (most of the state), KC and STL. • Unlike in other states where there is more than one system, in Missouri the three systems are completely separated (because of Social Security). • It is like having three different state systems all in one state, which is convenient because state-level administrative data from Missouri alone is sufficient to analyze mobility.
Data • 18-year administrative data panel from the entire state of Missouri • We can track within-state movement for all education personnel. • Size and length of data panel is important for leadership analysis. • We combine full and assistant principals for most of our analysis but separate out full principals in some cases. • In total, we observe over 11,000 leadership hires, approximately half of which are first time hires. Of all first time hires, 67 percent come directly from teaching in Missouri.
Initial Investigation of Across District Mobility • Suggestive evidence that mobility is limited by the pension borders. • But inference is clouded by non-comparability in the data (size and district disadvantage). • We take two approaches to isolate the pension-border effects • Simulations (omitted for time) • Simple models of leadership mobility between individual pairs of schools • Both approaches attempt to circumvent non-comparability between the groups of schools on different sides of the border through disaggregation.
Pairwise models • Break down the border-region data into individual pairs of schools and evaluate leadership sharing between all possible pairs. • In the KC region there are 16,836 unique pairs of schools; in STL there are 35,245 pairs. • Benefits: No re-sampling or extrapolation of findings required. • Primary Cost: Most pairs of schools do not share any hires – outcome is mostly populated with zeros. Therefore, we reduce the outcome variable to be binary and estimate a logit model. All instances of multiple shared hires between schools are dropped (26 percent of all shared hires in KC, 18 percent in STL).
Pairwise Results • Using the logit estimates, we predict estimate that removing the pension border between any two schools will increase the probability of a shared leadership hire by 85 percent in the KC region, and 96 percent in the STL region. • The simulations suggest a larger border effect in KC – we show that a main reason is that we cannot include information about multiple shared hires between schools in the pairwise analysis (limited data).
Concluding Remarks • We quantify pension-border effects on leadership mobility across schools. Our analysis suggests that pension borders meaningfully impact mobility. • Specific to Missouri, our findings suggest that Kansas City and St. Louis will have great difficulty in implementing any reforms that involve improving leadership quality. • General policy prescriptions to reduce the mobility tariffs of the current retirement structure in education include selective intervention by state governments, pension-system integration (unlikely), and/or moves toward retirement plans that do not penalize mobility (i.e., defined contribution or cash-balance plans).