Comparable, Schmomparable

download Comparable, Schmomparable

of 32

Transcript of Comparable, Schmomparable

  • 8/7/2019 Comparable, Schmomparable

    1/32

  • 8/7/2019 Comparable, Schmomparable

    2/32

  • 8/7/2019 Comparable, Schmomparable

    3/32

    Comparable, SchmomparableEvidence o Inequity in the Allocation o Funds orTeacher Salary Within Cali ornias Public School Districts

    Raegen T. Miller May 2010

  • 8/7/2019 Comparable, Schmomparable

    4/32

  • 8/7/2019 Comparable, Schmomparable

    5/32

    1 Executive summary

    3 Background

    8 Why study California?

    10 Data analysis

    13 Findings

    16 Discussion and conclusion

    18 Appendix

    22 Acknowledgements

    23 Endnotes

    Contents

  • 8/7/2019 Comparable, Schmomparable

    6/32

  • 8/7/2019 Comparable, Schmomparable

    7/32

    Exe u ve ummary | www.amer a pr gre

    Executive summary

    Inequi y haun s U.S. public school nance. Some ederal programs are demon-s rably un air in alloca ing unds o s a es, and here prevails in many s a es anega ive rela ionship be ween he ra e o s uden pover y in school dis ric s and

    he amoun o per s uden revenues made available by he s a e unding ormula.Tere is also reason o believe ha he dis ribu ion o unds o schools wi hindis ric s sys ema ically dis avors schools serving he highes concen ra ions o low-income s uden s. Te reason is ha unds ollow eacher experience. eacher

    salary, he larges ca egory o school expendi ure, is igh ly linked o seniori y,which also con ers rans er privileges. eachers end o exercise hese privileges o

    ee high-pover y schools or ones serving more a uen communi ies.

    Te empirical li era ure documen ing he ex en o wi hin-dis ric inequi y isas onishingly hin. Te da a necessary o assess such inequi yac ual expendi-

    ures a he school levelhave been almos comple ely absen rom he pic urehis orically. School dis ric s generally alloca e unds o schools using abs rac ,non nancial erms such as he ra io o s uden s o eachers, and dis ric s, noschools, pay eachers salaries. School budge s and expendi ure repor ing ail ore ec ac ual eachers salaries, which one expec s o be lower, on average, in high-pover y schools where eacher urnover holds down he average level o experi-ence. Tus, school dis ric s ordinary business prac ices can conceal salary gaps:di erences be ween he average salary o eachers in high-pover y schools and heaverage salary o eachers in low-pover y schools.

    In repid researchers and advoca es have made progress in pegging hidden salary gaps over he pas ew years. Laboriously cons ruc ed es ima es o schools average

    eacher salary reveal pervasive salary gaps among large school dis ric s in several

    s a es. Ye wi hin-dis ric inequi y is o po en ial concern in mos school dis ric sin every s a e. For una ely, he American Recovery and Reinves men Ac o 2009,also known as he s imulus bill, included a repor ing requiremen ha shouldenable researchers and advoca es o expose hidden salary gaps ar and wide.

  • 8/7/2019 Comparable, Schmomparable

    8/32

    2 ce er r Amer a Pr gre | c mparable, s m mparable

    Tis paper exploi s a unique da ase con aining in orma ion on a represensample o 1,692 Cali ornia public schools. Mos o he da a were drawn cen ralized les main ained by he Cali ornia Depar men o Educa ion aNa ional Cen er or Educa ion S a is ics, bu school-level average eachries were plucked, one by one, rom online school accoun abili y repor car

    Foreshadowing he repor ing requiremen o he s imulus bill, Cali ornia Bill 687 required schools o pos ac ual expendi ure da a, including averageacher salary, on hese elec ronic documen s.

    Analyses o his da a shed ligh on he ex en o wi hin-dis ric inequiCali ornia. A 10 percen increase in he ra e o s uden pover y in a Calipublic school is associa ed wi h a $411 drop in average eacher salary, on avage, con rolling or several charac eris ics o dis ric s and schools known

    unding s reams. Tis abs rac nding ransla es o concre e dispari ies inavailable o suppor ins ruc ion. Te aggrega ed salary gap be ween wo o

    iden ical schools wi h he average number o eachers, one wi h a s udenra e o 50 percen age poin s higher han he o her, amoun s o approxim$76,000. Fur her analyses demons ra e ha resul s are robus o a numbesi ivi y es s, and hey provide evidence consis en wi h he no ion haby which unds ollow experience are responsible or inequi y. Te magni upervasiveness o predic ed salary gaps corrobora es exis ing evidence romCali ornia dis ric s while poin ing o a s a ewide problem.

    Tis paper demons ra es me hods sui able or assessing salary gaps in a airsimple, and general way. I s ndings, while building knowledge and highlighCali ornias leadership in promo ing ransparency in school expendi ures, resen he ip o he iceberg o an underexamined ace o scal equi y. Tbill repor ing requiremen will enable researchers and advoca es o uncoverden salary gaps in o her s a es.

    Uncovering hese hidden gaps should be a high priori y or wo reasons. Firschool dis ric s wishing o alloca e resources in ways ha promo e s udeachievemen generally and help close achievemen gaps be ween low-incoms uden s and heir more a uen peers need a beter grip on how hey alloca

    resources in he rs place. Second, Congress needs a beter unders anding oinequi y curren ly condoned by a loophole in he comparabili y requiremeno hree scal requiremen s placed on dis ric s receiving unds under i lA o he Elemen ary and Secondary Educa ion Ac .

  • 8/7/2019 Comparable, Schmomparable

    9/32

    Ba kgr u d | www.amer a pr gre

    Background

    Charac erizing equi y in U.S. public school nance was once a simple mater.Schools in poor areas were poorly unded, and schools in weal hy areas were well

    unded. Vir ually, all revenues derived rom local axes levied on real es a e. Tesi ua ion oday is vas ly more complex in wo ways. Firs , he overall propor iono all school revenues rom local sources, s ill 80 percen in 1930,1 now hovers

    rom year o year around 44 percen . Tis propor ion was 43.9 percen in he2006-07 school year, wi h s a e and ederal revenues accoun ing or 47.6 and

    8.5 percen o he o al, respec ively.2 Second, urbaniza ion and adminis ra iveconsolida ion in he la e 19 h and early 20 h cen ury made he school dis ric ,no he school, he ocal poin o revenue policies and he agen responsible ordis ribu ing resources o schools.

    Whe her schools are equi ably unded depends no only on he dis ribu iono local, s a e, and ederal unds, bu also on resource alloca ion prac iceswi hin school dis ric s. Federal unds ow almos exclusively according o heneed-based ormulas o large programs, mos impor an ly i le I, Par A o heElemen ary and Secondary Educa ion Ac , commonly called i le I, which chan-nels unds o school dis ric s o enhance he educa ional experience o childrenliving in areas o concen ra ed pover y. Tere are legi ima e concerns abouequi y in he alloca ion o hese unds,3 bu he grea majori y o school nancere orm e or s have aimed o improve equi y in he dis ribu ion o non ederal

    unds be ween dis ric s wi hin s a es.

    Between-district equity

    S a e unding ormulas end o exer an equalizing e ec on per pupil rev-enues be ween dis ric s, on average, and no by acciden . Tese ormulas weresculp ed by wo genera ions o li iga ion and legisla ion seeking equi able oradequa e unding or proper y-poor school dis ric s.4 In some s a es, no ably

  • 8/7/2019 Comparable, Schmomparable

    10/32

    4 ce er r Amer a Pr gre | c mparable, s m mparable

    New Jersey, s a e revenues accruing o school dis ric s overcome disparilocal weal h o crea e a s rong posi ive rela ionship be ween combined local revenues available o a school dis ric and he percen age o i s sliving in pover y.5

    In o her s a es he rela ionship be ween school dis ric s non ederal reand heir pover y ra es is nega ive. Te legal s a us o hese s a es unormulas remains in ux accordingly. Te Connec icu Supreme Cour , or

    example, recen ly paved he way or a challenge o Connec icu s schoosys em by reversing a 2007 lower cour decision dismissing he cons i ubasis or a sui brough by a consor ium o low-income dis ric s, ci iesen s.6 Te Connec icu legisla ure could po en ially preemp a sui and saadvoca es or low-income s uden s by improving be ween-dis ric equisuch a move, while represen ing progress, would no address inequi y in s

    unding wi hin dis ric s.

    Within-district equity

    Scandalous inequi y in he dis ribu ion o resources wi hin school dis rihas plagued U.S. educa ion or more han a hundred years. Te persis ence

    hese ques ions is no or lack o e or s o address he underlying probEqui able resource dis ribu ion was a cen ral in eres , or example, o ncour -moni ored desegrega ion plans or decades.7 Similarly, dis ric s receiv-ing ederal unds under i le I are required o provide comparable serviaverage, in bo h heir schools serving concen ra ions o low-income s udand heir o her schools.8

    Te problem is ha schools dis ric s have managed o appear equi able in o cour moni ors or Depar men o Educa ion audi ors wi hou necessaso. Te main reason is ha compliance regimes do no ocus on ac ual expedi ures. Tey ocus ins ead on abs rac quan i ies such as he ra io o boos uden s or he ra io o s uden s o s a .9 Such quan i ies can be reasonably similacross schools even while ac ual per pupil expendi ures vary enormously.

    Tis s a e o a airs is convenien or school dis ric o cials, who ypicallyca e nancial resources, o a large ex en , in non nancial erms. Schools re

    slo s or eachers, adminis ra ors, and suppor personnel based primarilynumber and charac eris ics o s uden s enrolled in hem, no ha his is

  • 8/7/2019 Comparable, Schmomparable

    11/32

    Ba kgr u d | www.amer a pr gre

    orward exercise. Class size limi s and o her policies can complica e matersconsiderably, bu a he same ime, o cials are ree o ignore he rami ca ions hisapproach can have on unding equi y.

    Funds follow experience

    eacher experience is he driving orce behind he dis ribu ion o ac ual nancialresources wi hin school dis ric s or hree reasons. Firs , eacher salary cons i u es

    he larges ca egory o school expendi ures.10 Second, eacher salary increases inreal erms wi h addi ional years o experience, on average. An applicable rule o

    humb is ha eachers who s ay in he same school dis ric or 30 years can expecheir salaries o double, afer accoun ing or in a ion. Tird, radi ional rans er

    policies privilege seniori y. A eachers abili y o rans er as desired be ween wodis ric schools increases wi h experience.

    Tese rules have clear implica ions or nancial equi y. eachers, no unlike o herkinds o workers, pre er o work a si es where heir jobs are percep ibly easier,holding all else equal. Tis pre erence does no avor schools serving concen ra-

    ions o low-income children.11 Research shows ha eachers ofen move away rom high-pover y schools, ei her by securing a rans er wi hin dis ric , or chang-

    ing dis ric s.12 A any given ime, eachers in low-pover y schools exhibi higherlevels o experience, on average, han eachers in high-pover y schools. Tis nd-ing is rue across and wi hin dis ric s.13

    One migh expec here o be moun ains o evidence showing ha eachers in low-pover y schools earn higher salaries han eachers in high-pover y schools, bu i

    urns ou ha he relevan da a have been hard o come by. Average eacher sala-ries a he school level do no appear in he Common Core o Da a, he reposi ory

    or annual collec ions o in orma ion abou public schools made by he Na ionalCen er or Educa ion S a is ics. Nor have s a e educa ional agencies his orically had such in orma ion, much less made i available.

    Te reason or he dear h o school-level in orma ion on ac ual eacher salary a

    he school level is ha repor ing obliga ions align wi h he s andard business prac-ice o alloca ing eaching slo s ins ead o unds. School dis ric s pay eachers

    salaries, and he dis ric average is repor ed as hough i per ained o individualschools, hus concealing di erences in ac ual school-level average salary driven by

    eacher experience.14

  • 8/7/2019 Comparable, Schmomparable

    12/32

    6 ce er r Amer a Pr gre | c mparable, s m mparable

    Hidden salary gaps

    Advoca es have begun o ge a handle on he magni ude and pervasivenessinequi y in ac ual expendi ures on eachers salaries by pains akingly maschool-level in orma ion on eacher experience o dis ric -level salary sca

    Educa ion rus , a nonpro research and advocacy rm, assessed he hiddesalary gap wi hin he 50 larges school dis ric s in exas and he 14 largesby es ima ing he average eacher salary in schools serving he highes anconcen ra ions o low-income s uden s. Te Educa ion rus Wes asseshidden salary gap wi hin he 50 larges school dis ric s in Cali ornia.15

    Tese s udies yielded wo common ndings. Firs , he overwhelming majoridis ric s examined had subs an ially lower average eacher salaries in heir pover y schools han in heir low-pover y schools. Second, hese hidden diences in average salary commonly opped $1,000, and some even opped $6,0

    Fur her analyses showed ha hese unding gaps ofen persis afer being cono a per pupil basis o accoun or varia ion in s uden - eacher ra io. Te s

    provide s rong evidence ha large dis ric s in Cali ornia, Ohio, and exas sless, on average, o pay eachers in schools serving concen ra ions o low-inchildren han hey do o pay eachers in schools serving more a uen s ude

    Funds ollow experience in small- and medium-size school dis ric s or hereasons hey do so in large dis ric s, so i is reasonable o imagine ha hiary gaps bligh many small- and medium-size dis ric s, oo. And since 95 po dis ric s receive i le I unds, ederal policymakers would bene romma ion speaking o he ex en and magni ude o hidden salary gaps acrosdis ric s han jus he larges ones in hree s a es. Te reason is ha hiddgaps represen evidence ha a known loophole in he i le I scal requiremundermines he compensa ory purpose o i le I unds.16

    Te me hodology pioneered by Educa ion rus and Educa ion rus Wesdoes no lend i sel well o charac erizing hidden salary gaps across nearlydis ric s in a s a e. I relies purely on wi hin-dis ric comparisons ha w

    or dis ric s wi h sizable numbers o schools, bu here are analy ic ram

    capable o working simul aneously wi h da a rom schools across all o a ric s, large, medium, and small. Te lack o appropria e da a has been he oo bringing such rameworks o bear on ques ions abou hidden salary gap

    qui e recen ly ha is.

  • 8/7/2019 Comparable, Schmomparable

    13/32

    Ba kgr u d | www.amer a pr gre

    New era of responsibility

    Te American Recovery and Reinves men Ac o 2009 con ained a repor ingrequiremen ha blows his radi ional obs acle ou o he wa er. Under heac , each s a e educa ional agency mus urnish he Depar men o Educa ion

    wi h school-by-school expendi ure da a or he 2008-09 school year by March31, 2010.17 Depar men o Educa ion guidance speci es ha expendi ures berepor ed in several ca egories. One o hese ca egories is eacher salary.18

    Advoca es or grea er equi y in school spending will soon have access o po en-ially enligh ening da a rom each s a e. Analysis o his da a should be a high

    priori y because he impending reau horiza ion o he Elemen ary and Secondary Educa ion Ac provides a window o oppor uni y or closing he comparabili y loophole. Te novel y o hese da a and he myriad ques ions o which hey may speak, however, presen a real danger o s irring up a cacophony o compe ing

    messages based on all manner o analyses.

    Advoca es and researchers can ake wo precau ions o ensure ha ederal policy-makers receive a clear signal abou wi hin-dis ric equi y. Te rs precau ion is

    o adop a laser-like ocus on varia ion in average eacher salary in he near erm.Te majori y o expendi ures go o eacher salary,19 and hidden salary gaps speak direc ly o a speci c aw in he law, a loophole in he comparabili y provision o

    i le I ha explici ly ignores he ac ha unds ollow experience. Te secondprecau ion is o rely on me hods sui able or charac erizing hidden salary gaps in a

    air, simple, and general way.

    Tis paper o ers a guide or observing hese precau ions. I leverages a uniqueda ase con aining in orma ion abou a large, represen a ive sample o Cali orniaschools. Tis paper pilo s an analy ic approach and charac erizes he hidden

    eacher salary gap in Cali ornia. I adds o he s ill small knowledge base aroundwi hin-dis ric equi y while an icipa ing a surge o work in his area. A echnicalappendix o ers researchers a de ailed rea men o me hodological issues.

  • 8/7/2019 Comparable, Schmomparable

    14/32

    8 ce er r Amer a Pr gre | c mparable, s m mparable

    Why study California?

    Tere are many reasons o inves iga e he hidden eacher salary gap in CaliForemos among hem is an unusualperhaps unique among s a espublirepor ing requiremen . Preceding he 2009 American Recovery and ReinvesAc s similar requiremen by our years, Cali ornia Sena e Bill 687 requireschools o include speci c expendi ure da a on heir school accoun abili ycards.20 Te bill explici ly called or he repor ing o average eacher salarieschool level.

    Te bills par icular ocus on expendi ures made sense in he wake o heWilliamsv. State o Cali orniacase, a class ac ion lawsui brough on behal o s udenparen s rom more han 46 schools based on Cali ornias cons i u ional resibili y o ensure s uden s receive basic educa ional oppor uni ies. Tis susetled by way o sweeping legisla ion promo ing equi y and ransparency ilic educa ion, and S.B. 687 essen ially correc ed an oversigh in his legisla21

    S udying Cali ornia has wo general advan ages rom a researchers s andFirs , ex an knowledge o hidden salary gaps among Cali ornias larges d

    ric s allows one o compare ndings o bols er credibili y o new ones. Toppor uni y is especially welcome because no prior work has sys ema icallexplored expendi ure da a drawn rom school accoun abili y repor cards.Cali ornias size ensures ha even a represen a ive sample o Cali ornias schools and dis ric s is large enough o suppor appropria e s a is ical e

    Te na ure o Cali ornias school unding sys em o ers an addi ional advans udying varia ion in average eacher salary. Hidden salary gaps may be mornounced in Cali ornia han in o her s a es. A combina ion o proper y a

    and s a e Supreme Cour decisions in he 1970s radically al ered school nCali ornia. In sum, school unding was leveled down. Wi h some no able ex

    ions, school dis ric s generally enjoy qui e similar per pupil revenues romand local sources. Resources available o pay eachers salaries vary less bedis ric s han would be he case in many s a es. Tis means ha charac e

  • 8/7/2019 Comparable, Schmomparable

    15/32

    W y udy cal r a? | www.amer a pr gre

    schools may play a grea er role rela ive o charac eris ics o dis ric s in explain-ing eachers sor ing behavior, heir endency o seek and ob ain he mos desir-able posi ions available o hem, ei her in heir curren dis ric or in ano her one.Moreover, eachers salaries accoun or a higher percen age o school expendi-

    ures in Cali ornia as compared o o her s a es because school dis ric s chose o

    increase class sizes ra her han lower salaries in response o downward rends inrevenue.22 Tis observa ion augmen s he policy relevance o hidden eacher sal-ary gaps in Cali ornia.

  • 8/7/2019 Comparable, Schmomparable

    16/32

    10 ce er r Amer a Pr gre | c mparable, s m mparable

    Data analysis

    Data sources and sampling procedures

    Te da a used in his s udy were drawn rom hree sources. Firs , repor ed aveacher salaries were ex rac ed, school by school, rom online versions o 20

    School Accoun abili y Repor Cards.23 Second, supplemen al in orma ion abouschools and dis ric s or he same year was drawn rom he Cali ornia Depao Educa ion.24 Tird, addi ional in orma ion on schools and dis ric s in 2006-2

    he mos recen year available, were drawn rom U.S. Depar men o EducaCommon Core o Da a.25

    Te wo later sources o da a include in orma ion or 9,198 schools clus er1,036 dis ric s. Average eacher salary gures, however, were only sough resen a ive sample o hese schools. Te ini ial sample includes all schools 84 dis ric s receiving basic aid under Cali ornias school nance sys em. Tedis ric s local proper y ax revenues su ce o mee or exceed he s a epupil unding requiremen , which varies according o a complex ormula. Baid dis ric s receive only ca egorical gran s rom he s a e,26 and hey serve some o

    he na ions weal hies communi ies.

    Te ini ial sample includes 20 percen o he schools rom he remaining disknown in he school nance con ex as revenue-limi dis ric s, bu hese dwere divided in o wo groups or sampling purposes. Te sample includes all sc

    rom a randomly chosen 20 percen o 1,025 dis ric s wi h ewer han 62 sincludes a randomly chosen 20 percen o he schools in he 11 dis ric s wi62 schoolsLos Angeles Uni ed, San Diego Uni ed, and o hers.

    Measures

    A nal analy ic sample was selec ed using cri eria rela ed o he measures o an assessmen o hidden eacher salary gaps. Approxima ely 5 percen o

    in he original sample were omited rom analyses because hey had missing o

  • 8/7/2019 Comparable, Schmomparable

    17/32

    Da a a aly | www.amer a pr gre .

    implausible values or average eacher salary. wo ypes o implausible valuesreared heir heads. Firs , in some dis ric s, all schools repor ed he same average

    eacher salary. Tis patern represen s a sys ema ic viola ion o he in en o herepor ing obliga ion, and such dis ric s can scarcely be expec ed o help shed lighon hidden salary gaps. Second, some schools repor ed inordina ely low or high

    average eacher salary such as $5,484 or $114,408, respec ively. Tese ypes o values may represen da a en ry errors, compu a ional errors, or some local mis-in erpre a ion o he repor ing requiremen . Spo checks o collec ive bargainingcon rac s yielded a smalles plausible value o $35,621 and a larges o $86,090.

    Addi ional measures used in his research include s a us and demographic indica-ors. Schools wi h missing or implausible values or some o hese indica ors were

    omited rom he nal analy ic sample, which includes 1,692 schools clus ered in220 dis ric s. By and large, revenue-limi dis ric s in he nal analy ic sample wereno s a is ically dis inguishable rom he popula ion o revenue-limi dis ric s

    based on observable s a us or demographic charac eris ics, a necessary condi-ion or drawing conclusions abou he popula ion o schools based on s a is ical

    rela ionships ound among schools in he analy ic sample.

    Analytic approach

    Analy ic me hods should be chosen o sui he goal o he research: o charac er-ize he hidden eacher salary gap in Cali ornia in a air, simple, and general way.Packaging his goal in o a speci c ques ion se s he ball in mo ion:

    What is the typical di erence between the average salary o teachers in oneschool versus another school, identical in many respects but serving students 10percentage points more likely to be fom a low-income amily?

    Mul iple regression echniques are well sui ed o addressing his ques ion usinghe da a a hand. Te echniques allow one o employ s a is ical con rols in service

    o airness. Selec ed resul s can be presen ed s raigh orwardly, and s a is ical es ssuppor generalizing ndings o he popula ion o Cali ornia schools. Addi ional

    analyses can explore he robus ness o ndings, sur ace he need or special pre-cau ion around in erpre a ion, and highligh analy ic concerns or u ure work.

    Figure 1 helps o illus ra e he ideas guiding his ocused explora ion o varia ionin average eacher salary. Te heigh o each bar indica es how many o he 1,692schools in he analy ic sample had average eacher salaries alling in he corre-

  • 8/7/2019 Comparable, Schmomparable

    18/32

    12 ce er r Amer a Pr gre | c mparable, s m mparable

    sponding range. I is cer ainly reasonable osuspec ha he dis ric a school happens resen helps explain which bin i s average sala

    alls in o. Basic aid dis ric s have grea er ncial resources han revenue-limi dis ric s, an

    Cali ornias school nance sys em rea s unielemen ary, and high school dis ric s a bi dieren ly, par ly because hey ace di eren c

    in areas such as ranspor a ion and ex racur-ricular ac ivi ies. Te Comparable Wage Index developed by he Na ional Cen er or EducaS a is ics provides a way o comparing dis r

    acing similar expendi ure pressures.

    Analyses should also respec he ac ha e

    ers canno sor hemselves among schools in jany old way. A hird-grade eacher may seek a

    rans er among elemen ary schools wi hin a dric , or she may ob ain a posi ion a an elemary school in ano her dis ric . She is generall

    no able, however, o move o a high school, ai is a simple mater o respec analy ically sucgrade-level cons rain s on eacher sor ing.

    Figure 1

    Frequency plot of average teacher salaries in 2007-08for a representative sample of 1,692 California schools

    500

    450

    400

    350

    300

    250

    200

    150

    100

    50

    0

    Number of schools

    $35-

    40k

    $40-

    45k

    $45-

    50k

    $50-

    55k

    $55-

    60k

    $60-

    65k

    $65-

    70k

    $70-

    75k

    $75-

    80k

    $80-

    85k

    $85-

    90k

    Average salary ranges

  • 8/7/2019 Comparable, Schmomparable

    19/32

    F d g | www.amer a pr gre .

    Findings

    Te ndings are no surprising: he higher he propor ion o low-income s u-den s served by a school, he lower he average salary o he schools eachers.Con rolling or selec charac eris ics o schools and dis ric s, each 10 percen agepoin increase in he propor ion o low-income s uden s served by a school cor-responds o a $411 drop in he average salary o he schools eachers. Tus, orexample, a 50 percen age poin di erence in he s uden pover y ra e be ween

    wo hypo he ical schools corresponds o $2,055 di erence in average salary.

    Mul iplying his salary gap by 37, he average number o eachers in each o Cali ornias schools, yields a gure o $76,035, a sum roughly equivalen o heaverage salary and bene s o an addi ional eacher.

    Tese resul s s and up o ve sensi ivi y analyses. Firs , one migh suspec haLos Angeles, home o 8 percen o Cali ornias public schools, may inordina ely a ec analyses, bu i does no . Second, one migh be concerned abou general-izing he nding o he whole s a e rom a sample in which basic aid dis ric s areover-represen ed. Analyses omiting basic aid dis ric s ea ure similar es ima es o

    he rela ionship be ween s uden pover y ra es and average eacher salary, in bo hmagni ude and s a is ical signi cance. Tird, he nding wi hs ands he inclusiono addi ional s a is ical con rols represen ing he ra io o s uden s o eachers ina school. Tus, he hidden salary gap is no explained by dis ric s pre erences orhabi s around he alloca ion o eaching slo s based on numbers o s uden s ina school. Four h, analyses omiting he hree special educa ion schools and 82al erna ive educa ion schools sugges ha hese schools do no drive he ndingsin any appreciable way. Finally, analyses excluding dis ric s wi h ewer han six schools in he analy ic sample yielded similar resul s.

    I is impor an o poin ou ha he rela ionship be ween s uden pover y andaverage eacher salary vanishes when a measure o eacher experience is added

    o regression models. Tis phenomenon bols ers he case ha he hidden salary gap is a produc o unds ollowing experience. One e ec o seniori y-based pay and eacher sor ing is a dis ribu ion o resources ha sys ema ically dis avorshigh-pover y schools.

  • 8/7/2019 Comparable, Schmomparable

    20/32

    14 ce er r Amer a Pr gre | c mparable, s m mparable

    In i s naked orm, his papers key nding iseasy o dismiss. A 10 percen age poin changin a schools s uden pover y ra e is hard opic ure, and $411 represen s he cos o aminor car repair. Ye wi hin he same dis ric

    he s uden pover y ra e can vary drama icbe ween schools, he basis or wo ways osi ua ing he nding.

    Worst-case scenario

    Tere are more han 600 dis ric s in Cali orniain which he di erence be ween he highes alowes pover y ra es o schools exceeds 10 pe

    cen age poin s. In a quar er o dis ric s, hemum di erence exceeds 47 percen age poin s.Figure 2 por rays he maximum predic ed salagaps by ocusing on elemen ary schools wi hinuni ed, revenue-limi dis ric s. Te maximumobserved di erence in pover y ra es among

    elemen ary schools wi hin he 68 such dis ric s in he analy ic sample rango 96 percen age poin s. For dis ric s wi h large ranges in his sense, he m

    predic ed salary gap represen s serious money$2,000, $3,000, or nearly $4,0

    Classical gap analysis

    Maximum predic ed salary gaps raise serious ques ions abou equi y, bu no illus ra e he pervasiveness o salary gaps wi hin dis ric s. A hypo

    ric wi h nine elemen ary schools wi h a very low s uden pover y ra eone wi h a high s uden pover y ra e would have a pronounced maximumgap bu no a pervasive one. Tis is one reason why previous descrip ive s uies o salary gaps in large dis ric s have ocused on average salary gaps be

    schools wi h s uden pover y ra es in he rs and our h quar iles on dis ribu ion o ha measure.27

    Figure 2

    Predicted maximum salary gaps corresponding todifferences in student poverty rates among elementaryschools within 68 unified, revenue-limit school districts

    Percentage point difference

    $0

    $500

    $1000

    $1500

    $2000

    $2500

    $3000

    $3500

    $4000

    $4500

    0 20 40 60 80 100 120

  • 8/7/2019 Comparable, Schmomparable

    21/32

    F d g | www.amer a pr gre .

    Figure 3 por rays he predic ed salary gapbe ween hypo he ical schools a he 25 h and75 h percen iles on he dis ribu ion o hemeasure o s uden pover y, again ocusingon sample elemen ary schools wi hin uni ed,

    revenue-limi dis ric s. Te di erence be weenhe 75 h and 25 h percen iles on s uden pov-er y ra es among elemen ary schools wi hin he68 such dis ric s in he analy ic sample ranges

    rom 0 o 65 percen age poin s. Te predic edsalary gap exceeds $1,000 or more han a hirdo hese dis ric s.

    Figure 3

    Predicted salary gaps between hypothetical schoolsat the 75th and 25th percentiles on student povertybased on observed rates among elementary schoolswithin 68 unified, revenue-limit school districts

    Percentage point difference

    $0

    $500

    $1000

    $1500

    $2000

    $2500

    $3000

    0 10 20 30 40 50 60

  • 8/7/2019 Comparable, Schmomparable

    22/32

    16 ce er r Amer a Pr gre | c mparable, s m mparable

    Discussion and conclusion

    Prior research has documen ed hidden salary gaps wi hin large dis ric s, bupaper provides evidence speaking o a s a ewide phenomenon. A Cali orniaschools s uden pover y ra e is a good predic or o he average salary o

    eachers. A 10 percen age poin increase in s uden pover y ra e is associa$411 drop in average eacher salary, on average, con rolling or several charac

    ics o dis ric s and schools.

    Si ua ing his air, simple, and general resul among dis ric s in he analsample makes he equi y implica ions o eacher salary gaps more palpablSubs an ial varia ion in s uden pover y ra es across schools wi hin dispoin s o hidden eacher salary gaps ha are bo h subs an ial and comm

    his sense, he resul s o his research are highly consis en wi h hose odescrip ive work by Educa ion rus and Educa ion rus Wes . A subsand pervasive hidden salary gap is no surprising, and i should be unders oa consequence o policies in which unds ollow experience. Tese policies cundermine he in en o compensa ory unding s reams wi hou he prcau ions. Federal policymakers should be especially concerned abou his,

    i le I, he larges school program opera ed by he Depar men o Educaalloca es compensa ory unds o high-pover y schools.

    Close the comparability loophole

    Widespread, hidden salary gaps sugges ha he curren i le I comparabii y requiremen condones inequi y. Te reason is ha he provision expliciexcludes rom comparabili y de ermina ions salary di eren ials based on

    experience. Removing his exclusion rom he law would go a long way owensuring ha high-pover y schools receive a air share o resources.28

  • 8/7/2019 Comparable, Schmomparable

    23/32

    D u a d lu | www.amer a pr gre .

    Promote transparency around expenditures

    Tis paper has modeled an analy ic approach o summarizing one s a es hiddeneacher salary gap. In o her s a es, however, sys ema ic rela ionships be ween s u-

    den pover y and average eacher salary a he school level remain hidden behind

    a veil o secrecy crea ed by de aul business prac ices. Opaci y in he dis ribu iono nancial resources o schools is inde ensible in a new era o responsibili y.Fur hermore, school dis ric s wishing o alloca e resources in ways ha improves uden achievemen generally and narrow achievemen gaps would do well ounders and rs how and where hey ac ually spend heir unds.29

    Cali ornia can ake dis inc pride as a leader in promo ing ransparency around hedis ribu ion o ac ual resources o schools. Tis paper has exploi ed his ranspar-ency o shed ligh on he ex en o inequi y crea ed by policies in which unds ollow experience, and curren ly condoned by he very ederal law mean o enhance he

    educa ional experience a orded children in areas o concen ra ed pover y. Tere isreason o believe ha expendi ure paterns dis avoring high-pover y schools ob ainin mos s a es and he Dis ric o Columbia, and advoca es or low-income s uden sshould avail hemselves o soon- o-be-released da a s emming rom he AmericanRecovery and Reinves men Ac o see whe her his is indeed he case.

  • 8/7/2019 Comparable, Schmomparable

    24/32

    18 ce er r Amer a Pr gre | c mparable, s m mparable

    Appendix

    able A1 gives a breakdown o he analy ic sample by numbers o dis ric s aschools o di eren ypes, as de ned by grade levels served. S a us indica odis ric and school ype have some bearing in e or s o predic he averagesalary a he school-level because hey a ec he amoun o resources avail

    eachers. Cali ornias school nance sys em rea s basic aid and revenue-limric s qui e di eren ly. Basic aid dis ric s receive only ca egorical gran s. C

    and nonca egorical unds are alloca ed di eren ly, oo, depending on whe h

    dis ric is an elemen ary, high school, or uni ed dis ric . eachers can generasor hemselves among schools serving he same grade levels. No e: Schools a ed by coun y o ces o educa ions are excluded en irely rom his research

    able A2 o ers descrip ive s a is ics on selec indica ors or 1,692 schools ly ic sample (192 schools rom basic aid dis ric s and 1,500 schools rom revlimi dis ric s) and he popula ion o 9,006 revenue-limi schools. Revenueschools in he analy ic sample are s a is ically indis inguishable rom he p

    o revenue-limi schools on mos indica ors.A high propor ion o schools rom basic aiddis ric s are included in he analy ic sample. Texcep ions are he ew schools wi h missing oimplausible values or average eacher salary. Tin orma ion, in conjunc ion wi h randomizain he sampling process, provides a reasonablebasis or generalizing ndings crea ed by tinghypo hesized regression models o da a o hpopula ion o Cali ornia schools.

    Te regression echniques used here beginwi h he hypo hesized model represen ed by Equa ion 1,whereAvesalij represen s he average eacher saary in schooli in dis ricj,Dj represen s a vec or

    Table a1

    Numbers of districts and schools in the analytic sample,by types defined by grade-levels served

    District type Basic aid Revenue limit Totals

    Elementary district 36 83 119

    High school district 5 12 17

    Unifed district 15 69 84

    Totals 56 164 220

    School type Basic aid Revenue limit Totals

    Elementary school 136 1009 1145

    Middle school 22 242 264

    High school 34 249 283

    Totals 192 1500 1692

    ijjijjij S DAvesal ++++=

  • 8/7/2019 Comparable, Schmomparable

    25/32

  • 8/7/2019 Comparable, Schmomparable

    26/32

  • 8/7/2019 Comparable, Schmomparable

    27/32

    Appe d x | www.amer a pr gre .

    In column (4), he e ec o an indica or o he average number o years o experi-ence o eachers in a school wipes ou he e ec o s uden pover y. Tis is nosurprising given he prevalence o policies by which unds ollow experience. Teinclusion o he experience predic or explains 34 percen o he wi hin dis ricvaria ion in average salary lef unexplained in column (3).

  • 8/7/2019 Comparable, Schmomparable

    28/32

    22 ce er r Amer a Pr gre | c mparable, s m mparable

    Acknowledgements

    Te au hor wishes o acknowledge he invaluable research assis ance o ErinPollard. Her collec ion, prepara ion, and preliminary analyses o he da a m

    his paper possible. Any errors in he nal analyses are due o he au hor.

    A note about the title

    Te main i le o his paper, Comparable, Schmomparable, was inspired by paper by Je rey B. Liebman and Richard J. Zeckhauser o Harvard Universi30 Teir paper has appeared in various i era ions wi h he i le Schmeduling,re ers o behavior o consumers presen ed wi h ex raordinarily complex pin orma ion. Te paper has no hing o do wi h school nance. Te i le, hownicely invokes he Yiddish conven ion o replacing an ini ial consonan wi

    schm o crea e a erm o derision.31

  • 8/7/2019 Comparable, Schmomparable

    29/32

    E d e | www.amer a pr gre .

    Endnotes

    1 t e ra l u d r m l al, a e, a d ederal ur evary a r a e . F r a e expla a , ee ap er 3, e 59 Er A. hu u ek a d Al red A. L d e ,Schoolhouses, Courthouses,and Statehouse :Solving the Funding-Achievement Puzzle in AmericasPublic Schools . (Pr e , nJ: Pr e U ver y Pre , 2009).

    2 U.s. Depar me Edu a , na al ce er r Edu a s a ,Per e age d r bu reve ue r publ eleme ary a d

    e dary edu a e U ed s a e , by ur e: 200607 (2009),ava lable a p:// e .ed.g v/EDFin/grap _ p .a p?inDEX=4.

    3 Raege M ller, se re Re pe Revealed: Demy y g e t lei, Par A Fu d g F rmula (Wa g : ce er r Amer aPr gre , 2009), ava lable a p://www.amer a pr gre . rg/

    ue /2009/08/ le_ e. ml.

    4 F r a e rea me e w ra l a e l ga-, ee hu u ek a d L d e ,Schoolhouses, Courthouses, and

    Statehouse :Solving the Funding-Achievement Puzzle in AmericasPublic Schools .

    5 G rd Ma i e ,In Plain Sight: Simple, Di cult Lessons rom New Jerseys Expensive Efort to Close the Achievement Gap (new Y rk: t ece ury F u da Pre , 2009).

    6 Ar elle Lev Be ker, s a e h g c ur Rul g c e need rQua y g Edu a Qual y,The Hart ord Courant , Mar 23,2010, ava lable a p://ar le . ura . m/2010-03-23/ ew /

    - l- u d g-law u -0323.ar mar23_1_ g er-edu a -adequa e-edu a -publ -edu a .

    7 M re a 200 d r rema u der e uperv e Depar -me Ju e w regard mpl a e w ur - rderedde egrega pla . F r a ere g urre example, ee

    s ep a e M crumme , Rul g ra al la M . s lre e r ubl g br ader re d,The Washington Post , Apr l 20,2010, ava lable a p://www.wa g p . m/wp-dy / -

    e /ar le/2010/04/19/AR2010041905118. ml.

    8 Elementary and Secondary Education Act, se 1120A( ) (1) (A). 20U.s.c. 6321. (G ver me Pr g o e, 2004).

    9 Depar me Edu a , n -Regula ry Gu da e: t le i F ali ue (2008), ava lable a p://ed.g v/pr gram / le par a/

    algu d.pd .

    10 t da e, edu a al expe d ure are w dely rep r ed a el level. Edu a Re ur e s ra eg e , a ul a y, pr v ded

    e au r w e ma e r e per e age l-levelexpe d ure due ea er alary a a d ul l e d r . E -

    ma e vary g a ly a r d r a d a aly al appr a e .t e m ua ed e ma e ra ge r m 42 per e 55 per e .t ev de e eem rea able der g a , a rd g

    d r -level da a lle ed by e na al ce er r Edu as a , 40 per e expe d ure are d re ed alar e -ru al per el (60 per e expe d ure are ded a ed

    ru , a d 67 per e a pe d g g e ward alar e ).see na al ce er r Edu a s a , F a e Grap , ava l-able a p:// e .ed.g v/EDFin/grap _ dex.a p.

    11 ce er r tea g Qual y, n r car l a tea er W rk gc d survey i er m Rep r (2006), ava lable a p://www.

    ea gqual y. rg/pd /2006 w er m.pd (la a e edo ber 28, 2008).

    12 Be jam s a d , Dav d L. sj qu , a d t dd R. s ebr k er, Dtea er Really Leave r h g er Pay g J b Al er a ve o upa-

    ?Advances in Economic Analysis & Policy 6 (1) (2006), ava lablea p://www. alder e er. rg/PDF/1001057_h g _P ver y.pd .

    13 c arle cl el er a d er , h g -P ver y s l a d e D r bu- tea er a d Pr pal . W rk g Paper 1 (Wa g : t e

    Urba i u e, ce er r e A aly L g ud al Da a Edu a , 2007).

    14 Marguer e R za, Larry M ller, a d Paul h ll, s re g e g t le i help h g -P ver y s l : h w t le i Fu d F i D r Al-l a Pa er (sea le: ce er Re ve g Publ Edu a ,2005); Marguer e R za, W a We cl ed e t le i c mparab l yL p le? i P lly M clure a d er , E ur g Equal opp r u-

    y Publ Edu a (Wa g : ce er r Amer a Pr gre ,2008); Kare hawley M le a d Marguer e R za, U der a d-

    g s ude -We g ed All a a a Mea Grea er s lRe ur e Equ y,Peabody Journal o Education 81 (3) (2006): 36-62.

    15 t e Edu a tru , t e r Fa r s are: h w tea er salary Gaps r a ge P r c ldre texa (2006), ava lable a p://www.ed ru . rg/ e /ed ru . rg/ le /publ a / le /texa %20(P r).pd ; t e Edu a tru We , cal r a h dde tea er spe d-

    g Gap: h w s a e a d D r Budge g Pra e s r a geP r a d M r y s ude a d t e r s l (2005), ava lable a

    p://www. dde gap. rg/re ur e /rep r 031105; t e Edu atru , n A u g r Fa r e : Equ able Edu a Fu d g(2008), ava lable a p://www.ed ru . rg/ e /ed ru . rg/ le /publ a / le /n A g rFa r e oh.pd .

    16 saba B reda a d Raege M ller, Walk g e talk: cl g ec mparab l y Requ reme L p le t le i e Eleme arya d se dary Edu a A (Wa g : ce er r Amer a

    Pr gre , 2010).

    17 U.s. Depar me Edu a , Amer a Re very a d Re ve -me A 2009: t le i, Par A Fu d r Gra L al Edu aAge e (2009), ava lable a p://ed.g v/p l y/ge /leg/re v-ery/ a ee / le- . ml.

    18 t e vel y rep r g requ reme may repre e a realb a le u e ul rep r g me a e . F r a example

    gu da e fered by a a e edu a al age y l d r ,ee s a e Wa g , o e super e de Publ

    i ru , s l App r me a d F a al serv e (2010),ava lable a p://www.k12.wa.u / a /bldg_lvl/ br.a p.

    19 see e 10.

    20 F r a de a led exam a e ge e , pa age, a d e r eme se a e B ll 687, ee J Afeld a d Gu llerm Mayer, L g e

    F g Average : E a g a d impleme g cal r a Requ re-

    me Rep r A ual Per Pup l Expe d ure s l-by-s l(Wa g : ce er r Amer a Pr gre , 2010), ava lable ap://www.amer a pr gre . rg/ ue /2010/05/l g e g. ml.

    t e b ll requ re l rep r a ual per pup l expe d ure( r m re r ed ur e , u re r ed ur e , a d all ur e ) a daverage ea er alary a par e l a u ab l y rep r

    ard. c mpl a e w e e rep r g requ reme , a pr r y ad-v a e e e ar ly l d r a d e cal r a Depar -me Edu a , r e qu kly u a 84 per e l were

    mpl a r e 2007-08 l year. Publ Adv a e , a law rm,

    http://ed.gov/programs/titleiparta/fiscalguid.pdfhttp://ed.gov/programs/titleiparta/fiscalguid.pdfhttp://nces.ed.gov/EDFIN/graph_index.asphttp://www.k12.wa.us/safs/bldg_lvl/sbr.asphttp://www.americanprogress.org/issues/2010/05/liftingthefog.htmlhttp://www.americanprogress.org/issues/2010/05/liftingthefog.htmlhttp://www.k12.wa.us/safs/bldg_lvl/sbr.asphttp://nces.ed.gov/EDFIN/graph_index.asphttp://ed.gov/programs/titleiparta/fiscalguid.pdfhttp://ed.gov/programs/titleiparta/fiscalguid.pdf
  • 8/7/2019 Comparable, Schmomparable

    30/32

    24 ce er r Amer a Pr gre | c mparable, s m mparable

    a d ume ed e aga mpl a e w s.B. 687, ava lablea p://www.publ adv a e . rg/d /sARc08/2008_sARc_Re-p r _Rev1_FinAL.pd .

    21 Publ Adv a e , i ., e e -lead u el eWilliamsv. State o Cali ornia a e, ma a a web e d ume g e

    mel e a d key eve ar u d e a e, ava lable a p://www.publ adv a e . rg/ urw rk/edu a / dex. ml#w ll am .

    22 Publ P l y i u e cal r a, ha s l F a e Re rm BeeG d r cal r a? Re ear br e umber 30 (2000), ava lable a

    p://www.pp . rg/ e /pub /rb/RB_200JsRB.pd .

    23 s l A u ab l y Rep r card , r sARc , are mea ludeur expe d ure mea ure : average ea er alary, al per pup l

    expe d ure , per pup l expe d ure r m re r ed ur e , a dper pup l expe d ure r m re r ed ur e . All ur valuewere lle ed, w ere ava lable. Expl ra ry a aly ugge a

    e per pup l expe d ure mea ure are le red ble a e aver-age ea er alary mea ure. A plau ble expla a a ere arem re way erpre expe d ure a way erpre ea er

    alary. D r average rea ed r m e l-level expe d uregure d d rrela e r gly w d r average expe d uregure ava lable el ew ere. i ra , m larly ru ed d rea er alary gure rrela ed r gly w mea ure ru edr m d r expe d ure da a ava lable el ew ere.

    24 t e cal r a Depar me Edu a pr v de a e var uda a e by way a a dy web e, ava lable a p://dq. de.

    a.g v/da aque /.i al ma a a e ral p r al a l a ga e sARc, ava lable a p://www. de. a.g v/ a/a / a/ap/

    ar l k1.a p.

    25 see U.s. Depar me Edu a , i u e r Edu a s ena al ce er r Edu a s a , c mm c re Da a,ava lable a p:// e .ed.g v/ d/.

    26 Eds ur e, a pr re ear rm, fer ma y u e ul d u-me de r b g cal r a l a e y em, lud g

    ll w g: Eds ur e, Ab u reve ue l m a d ba a d (2004),ava lable a p://r d. lw re . e /187710123165237197/l b/187710123165237197/Ab u _reve ue_l m _a d_ba _a d.pd .

    27 t e Edu a tru me d l gy ra k l by p ver y a de d v de em ur gr up erv g equal umber ude . i d e alway arra ge r mpar by grade ba d

    28 see B reda a d M ller, Walk g e talk: cl g e c mparab l yRequ reme L p le t le i e Eleme ary a d se daryEdu a A .

    29 Kare hawley M le a d s ep e Fra k,The Strategic School: Makingthe Most o People, Time, and Money (t u a d oak , cA: c rwPre , 2008).

    30 see, r example, a 2004 ver Jefrey B. L eberma a d R ardJ. Ze k au er, s medul g (cambr dge, MA: harvard U ver ya d e na al Bureau E m Re ear , 2004), ava lable a

    p://www.e .yale.edu/~ ller/be ma r /2004-11/ medul-g-ze k au er.pd .

    31 see, r example, Merr am-Web er o l e d ary, ava lable ap://www.merr am-web er. m/d ary/ m.

    http://www.publicadvocates.org/ourwork/education/index.html#williamshttp://www.publicadvocates.org/ourwork/education/index.html#williamshttp://www.ppic.org/content/pubs/rb/RB_200JSRB.pdfhttp://dq.cde.ca.gov/dataquest/http://dq.cde.ca.gov/dataquest/http://www.cde.ca.gov/ta/ac/sa/ap/sarclink1.asphttp://www.cde.ca.gov/ta/ac/sa/ap/sarclink1.asphttp://nces.ed.gov/ccd/http://rcsd.schoolwires.net/187710123165237197/lib/187710123165237197/About_revenue_limits_and_basic_aid.pdfhttp://rcsd.schoolwires.net/187710123165237197/lib/187710123165237197/About_revenue_limits_and_basic_aid.pdfhttp://www.econ.yale.edu/~shiller/behmacro/2004-11/schmeduling-zeckhauser.pdfhttp://www.econ.yale.edu/~shiller/behmacro/2004-11/schmeduling-zeckhauser.pdfhttp://www.merriam-webster.com/dictionary/schmhttp://www.merriam-webster.com/dictionary/schmhttp://www.econ.yale.edu/~shiller/behmacro/2004-11/schmeduling-zeckhauser.pdfhttp://www.econ.yale.edu/~shiller/behmacro/2004-11/schmeduling-zeckhauser.pdfhttp://rcsd.schoolwires.net/187710123165237197/lib/187710123165237197/About_revenue_limits_and_basic_aid.pdfhttp://rcsd.schoolwires.net/187710123165237197/lib/187710123165237197/About_revenue_limits_and_basic_aid.pdfhttp://nces.ed.gov/ccd/http://www.cde.ca.gov/ta/ac/sa/ap/sarclink1.asphttp://www.cde.ca.gov/ta/ac/sa/ap/sarclink1.asphttp://dq.cde.ca.gov/dataquest/http://dq.cde.ca.gov/dataquest/http://www.ppic.org/content/pubs/rb/RB_200JSRB.pdfhttp://www.publicadvocates.org/ourwork/education/index.html#williamshttp://www.publicadvocates.org/ourwork/education/index.html#williams
  • 8/7/2019 Comparable, Schmomparable

    31/32

  • 8/7/2019 Comparable, Schmomparable

    32/32