D I S C U S S I O N P A P E R S E R I E S IZA DP No. 6168 Homeownershp, Socal Captal and Parental Voce n Schoolng Arthur Grmes Steven Stllman Chrs Young November 2011 Forschungsnsttut zur Zukunft der Arbet Insttute for the Study of Labor
Homeownershp, Socal Captal and Parental Voce n Schoolng Arthur Grmes Motu Economc and Publc Polcy Research and Unversty of Auckland Steven Stllman Unversty of Otago and IZA Chrs Young Motu Economc and Publc Polcy Research Dscusson Paper No. 6168 November 2011 IZA P.O. Box 7240 53072 Bonn Germany Phone: +49-228-3894-0 Fax: +49-228-3894-180 E-mal: za@za.org Any opnons expressed here are those of the author(s) and not those of IZA. Research publshed n ths seres may nclude vews on polcy, but the nsttute tself takes no nsttutonal polcy postons. The Insttute for the Study of Labor (IZA) n Bonn s a local and vrtual nternatonal research center and a place of communcaton between scence, poltcs and busness. IZA s an ndependent nonproft organzaton supported by Deutsche Post Foundaton. The center s assocated wth the Unversty of Bonn and offers a stmulatng research envronment through ts nternatonal network, workshops and conferences, data servce, project support, research vsts and doctoral program. IZA engages n () orgnal and nternatonally compettve research n all felds of labor economcs, () development of polcy concepts, and () dssemnaton of research results and concepts to the nterested publc. IZA Dscusson Papers often represent prelmnary work and are crculated to encourage dscusson. Ctaton of such a paper should account for ts provsonal character. A revsed verson may be avalable drectly from the author.
IZA Dscusson Paper No. 6168 November 2011 ABSTRACT Homeownershp, Socal Captal and Parental Voce n Schoolng * We use New Zealand school board of trustees data to examne whether schools where parents have hgh rates of homeownershp experence hgh parental votng turnout n electons. We also nvestgate whether homeownershp nfluences the probablty that a school board proceeds to electon, ndcatng parental wllngness to serve as a school trustee. Smlarly, we examne whether state-owned socal housng rates affect these outcomes. We comple results ntally wthout controllng for other factors, and then controllng for a wde range of other characterstcs, to test the robustness of smple observed assocatons between homeownershp and state-ownershp rates and outcome varables. Our fndngs show no dscernble effect of homeownershp on parental votng turnout n school electons after controls are added (contrary to the smple postve assocaton), but a (robust) postve mpact of both homeownershp and state-ownershp rates on the probablty that a school holds an electon. JEL Classfcaton: I28, R23, Z13 Keywords: homeownershp, school electons, parental voce, socal captal Correspondng author: Steven Stllman Department of Economcs Unversty of Otago PO Box 56 Dunedn 9054 New Zealand E-mal: steven.stllman@otago.ac.nz * We thank semnar partcpants at the Unversty of Otago and the Swss Sk-Labor Economcs Workshop for ther useful comments. We also thank the Marsden Fund of the Royal Socety of New Zealand (grant: 07-MEP-003, Homeownershp and Neghbourhood Wellbeng) for provdng the fnancal assstance for ths research and the New Zealand Mnstry of Educaton for provdng data for ths study. The authors reman solely responsble for the analyss and the vews expressed.
1. Introducton The effects of homeownershp on socal and economc outcomes are wdely dscussed. Studes have noted benefts of homeownershp on outcomes such as: wealth; labour force partcpaton; urban structure; health; demographcs; self-esteem; chld outcomes; and socal captal outcomes such as nvolvement n poltcal and socal actvtes (Detz and Haurn, 2003). In ths paper, we focus on one specfc socal captal (chld-related) outcome that s potentally affected by homeownershp: parental votng turnout n local school board of trustees electons. 1 Reflectng common perceptons of homeownershp effects, we hypothesse that schools n areas wth hgher rates of homeownershp wll experence a hgher parental votng turnout n board of trustees electons (relatve to a null of no effect). A school s a form of local amenty and the qualty of amentes affects the values of propertes surroundng them. Homeownng parents have a greater ncentve than renters to vote n board of trustees electons, as the board s responsble for the operaton and performance of the school. If a school underperforms, ths can generate an unfavourable externalty on local house prces whch drectly affects homeowners, but not renters (ndeed, rents may declne). In addton, homeowners are less moble than renters so there s a greater cost to them f they were to shft n response to school qualty, agan ncentvsng homeowners to act (more than renters) to rase school qualty. Studes of the mpact of homeownershp on local outcomes have to account for the mpacts of covarates that are correlated wth homeownershp and that themselves may mpact on outcomes of nterest. Ths s partcularly mportant gven that homeownershp s commonly assocated wth other markers of soco-economc status such as educaton and 1 The concept of socal captal was ntally formulated wth reference to the role of schools n the local communty (Hanfan, 1916). 1
ncome. We pay partcular attenton to controllng for other factors and demonstrate the mportance of dong so relatve to an approach that reles on smple correlatons between homeownershp and outcome varables. Our results ndcate that there s a postve assocaton between homeownershp and parental votng turnout n school electons. However, once school and local characterstcs are controlled for, there s no dscernble effect of homeownershp on voter turnout. In addton to nvestgatng the effect of homeownershp on votng rates, we analyse whether homeownershp rates affect the probablty of a school proceedng to an electon. In 2007, only around half of prmary and ntermedate schools proceeded to an electon, whereas 75% of secondary schools held an electon. Schools proceed to an electon where there are more canddates than avalable board postons; thus havng an electon s a marker of parental wllngness to actvely serve the school n the role of (unpad) trustee. We fnd that hgh homeownershp does ncrease the chance of a prmary school proceedng to electon. Smlarly, the probablty of an electon ncreases wth a hgher proporton of state-owned houses, consstent wth the theory that housng stablty (through ether homeownershp or state tenancy) leads to a greater sense of communty relatve to other forms of tenure. The paper s structured as follows: Secton 2 presents a bref revew of prevous lterature; Secton 3 gves a background to the New Zealand schoolng system; our methodology s outlned n Secton 4; data used n the study s descrbed n Secton 5; Secton 6 provdes results of estmaton; and conclusons are provded n Secton 7. 2. Pror Lterature Governments n many countres formulate polces and strateges to encourage hgher homeownershp rates as evdence suggests that there are postve externaltes assocated wth homeownershp. In the U.S., Government polcy (usng tax breaks and subsdes) has consstently been drected toward encouragng ctzens to become homeowners (Haurn et al., 2
2002; Green and Whte, 1997). In New Zealand, mputed rents for owner-occupers reman tax-free (Tax Workng Group, 2010). One fundamental dfference between homeownershp and rentng s the transacton cost assocated wth securng and vacatng a dwellng (Detz and Haurn, 2003). Transacton costs are sgnfcantly greater for homeowners than for renters and, as a result, homeowners are lkely to be less moble, or more geographcally stable, than renters. Also, the decson to own or to rent s a matter of choce that s heavly constraned and dependent on the type of dwellng a household desres to resde n (Ross and Weber, 1996). For example, n the U.S., f a household wants to lve n a sngle-famly detached house, they are largely confned to the ownershp market as few such unts are avalable for rent. The benefts of ownng a home can be categorsed nto three man types of personal and/or neghbourhood beneft (Boehm and Schlottmann, 1999). Frst, for many famles, homeownershp s the largest nvestment they wll ever make. Therefore, the home s an asset whch wll provde future fnancal securty for those famles who can afford to purchase. Second, homeowners are able to gan a hgher level of personal esteem and lfe satsfacton, whch develops them and ther chldren nto more productve members of socety. Thrd, reduced moblty of homeowners helps mprove neghbourhood qualty and stablty. Reduced moblty may ncentvse homeowners to be more soco-poltcally actve than renters n order to mprove the neghbourhood envronment, whch s then captalsed nto property values. Koff and Sen (2005) observed the effects of homeownershp on cvc effort. They argue that sustaned cvc efforts lead to mprovements n the local envronment, whch are then captalsed n property values. Cvc efforts, however, are not contractble, so there needs to be some ncentve for households to exert more effort. As homeowners gan from ncreases n property values, homeownershp creates an ncentve for homeowners to exert greater cvc 3
effort (whch ncludes votng n local electons) to mprove the qualty of ther propertes and communty n order to rase local property values. There are very few studes that nvestgate the specfc relatonshp between school votng turnout and homeownershp. However, studes have nvestgated the homeownershp mpacts on local body electons and poltcal actvty, cvc effort and neghbourhood partcpaton. Homeowners are more ncentvsed than renters to be poltcally and socally nvolved n local affars. Homeownershp nfluences socal behavour through two channels frst, homeownershp s an nvestment that alters the fnancal stake of households. Local affars can nfluence house prces and, subsequently, the value of the nvestment. Homeowners therefore have an added ncentve to be actve n local affars (Detz and Haurn, 2003; Manturuk et al, 2010). Second, homeownershp reduces the moblty of households, whle rentng households are relatvely moble. The reduced moblty ncentvses homeowners to mantan and mprove the neghbourhood s qualty-of-lfe, as ths s drectly related to ther own qualty-of-lfe and s also captalsed nto the value of propertes wthn that neghbourhood. Hence, homeowners are more lkely to partcpate n poltcal and communty actvtes, and ncrease cvc efforts, to ensure any negatve externaltes to ther neghbourhood mage are mnmsed (Ross and Weber, 1996; Dpasquale and Glaeser, 1999; Detz and Haurn, 2003; Koff and Sen, 2005; Manturuk et al, 2010). Homeowners are consstently found to vote at greater rates than renters; however the effect of homeownershp on poltcal nterest s neglgble (Ross and Weber, 1996; Detz and Haurn, 2003). Dpasquale and Glaeser (1999) fnd that homeowners n the Unted States are 15% more lkely to vote n local electons than renters and that 77% of owners had sad they voted n local electons, compared wth only 52% of renters. When repeatng the analyss usng German data, the authors found much the same behavours, wth homeownershp 4
ncreasng the probablty of votng for low-ncome ndvduals by 4.3%, and hgh-ncome ndvduals by 29.1%. Recent evdence on the mpact of homeownershp on votng rates s, however, less clear-cut. Whle Manturuk et al (2010) fnd that homeownershp rates have a sgnfcant postve effect on local poltcal partcpaton, Engelhardt et al (2010) fnd no evdence that such a relatonshp exsts. Ths lack of certanty means that the mpact of homeownershp on electoral partcpaton remans an open queston. We examne ths ssue usng a tghtly defned form of electoral nvolvement: votng n one s own chld s school board of trustees electon. Compared wth other forms of votng, the school electon s closely ted to an outcome of drect personal nterest and so provdes a sharper test of homeownershp versus renter propensty to be poltcally nvolved than pror studes where benefts of votng are more dffused. 3. New Zealand Schools Background 3.1. School Governance and Structure The governance of New Zealand schools can be broadly categorsed nto State, State- Integrated and Other. State schools receve government fundng (that can be supplemented by voluntary donatons). They are generally co-educatonal2 and all are requred to teach the New Zealand currculum. State-ntegrated schools were prevously prvate schools, but are now part of the state system. They ncorporate ther own specal character (usually a phlosophcal or relgous belef) nto the New Zealand currculum that they teach. Ther buldngs and property are prvately owned, but they receve the same fundng per student as state schools. In addton, they may charge compulsory fees to meet ther property costs. 2 Some secondary schools offer sngle-sex educaton. Of the state secondary schools 13.5% are boys only, 16.7% are grls only, and 69.8% are co-educatonal. Of the number of students attendng state schools 14.4% attend boys only schools, 15.1% attend grls only schools and 70.5% attend co-educatonal schools. 5
Prvate, or ndependent, schools stll receve some fundng from the Government, but the majorty of ther fundng s receved from the fees they charge. They have ther own ndependent boards whch govern them and must meet certan standards to be regstered wth the Mnstry of Educaton. Whle they are free to teach ther own currculum, t must follow a learnng program smlar to that of the New Zealand currculum. There s also a range of other types of schools that cater for the specfc needs of ther students (Mnstry of Educaton, 2009a). Table 1 provdes the number of schools and pupls by school governance and type. Table 1 shows that a large majorty (85%) of students n New Zealand attend state schools. The structures of schools n New Zealand are determned by the age of students they enrol. Prmary school s generally consdered to encompass students n Year 1 to Year 8, whle secondary schools encompass students n Year 9 to Year 13. The two man types of prmary schools operatng n New Zealand are contrbutng prmary and full prmary schools. Full prmary schools enrol students from Year 1 through to Year 8, whle contrbutng prmary schools only enrol students from Year 1 through to Year 6. Intermedate schools complement contrbutng schools by teachng only students n Year 7 and Year 8. Generally, secondary schools n New Zealand enrol students from Year 9 through to Year 13, however other secondary school structures exst. 3 There are also a number of other less common school age structures that operate (Mnstry of Educaton, 2009b). 4 3.2. Tomorrow s Schools Reforms and Boards of Trustees The Tomorrow s Schools reform process, mplemented n 1989, comprsed a number of structural changes n the New Zealand state educaton system that altered the way New 3 Some secondary schools also nclude the ntermedate years (.e. Year 7 through to Year 13). 4 Other school structures nclude: Area/composte schools (Year 1 13); Mddle schools or junor hgh schools (Years 7 10); senor hgh schools (Years 11 13); and another type of composte school (Years 1 10). 6
Zealand state schools were admnstered (Fske and Ladd, 2000; Mnstry of Educaton, 2009c). The domnant feature of the reforms was the transfer of the responsblty for runnng a school from the Department of Educaton to a locally elected board of trustees (BOT). The purpose of the BOT was to oversee the governance of a school. Its major responsbltes ncluded: the management of fnances; establshment and montorng of the school s strategc objectves; mantenance of physcal property; and employment of staff ncludng the prncpal. The Government also requred assurance from the BOTs that students would receve a hgh qualty standard of educaton and that resources made avalable to schools would be used effcently and effectvely (Mnstry of Educaton, 2009d). Each New Zealand state and state-ntegrated school s requred to have a BOT (Educaton Act, 1989). Electons for BOT representatves are held trennally, wth the frst electon held n 1989. 5 The standard consttuton of a board s three to seven parent-elected representatves, 6,7 the prncpal, one staff-elected representatve, a student-elected representatve 8, co-opted trustees, and up to four propretors representatves 9. There s a degree of flexblty wthn the composton of the board; however before changes can be made, the current board must gve the parent communty reasonable notce of ts ntentons (Mnstry of Educaton, 2009e). Parent representatves are elected by the parent communty and/or adult students of the school. The staff representatve s elected by staff members. Students are able to vote f the school qualfes for a student representatve. Electons for student representatves must be held annually n September. The board tself can also co-opt addtonal trustees; however, the number of co-opted trustees cannot exceed the number of 5 Electons are usually held on the second Tuesday n May of the electon year, unless the school s a correspondence school or the board, before February 1, has fxed an earler date for the electon. 6 A parent refers to a person who s the chld s father, mother, guardan, or mmedate caregver. 7 Orgnally, these seats were reserved for parent only, but, as part of the Educaton Amendment Act 1992, a provson was made that gave people wthout chldren attendng a gven school the rght to put themselves forward for electon. 8 A student representatve s elected only n schools wth Year 9 students or above. 9 Propretor s representatves are only apponted n ntegrated schools. They are usually elected to assure that the phlosophcal or relgous belef of the ntegrated school s upheld n BOT decson makng. 7
parent trustees. In the case where a vacancy for a trustee arses, the board s able to hold a byelecton to fll the vacancy; however, a co-opted trustee can only be apponted f another coopted poston becomes avalable (Educaton Act, 1989). A md-term, or staggered, electon cycle can also be adopted by the board, where half of the parent representatves are elected halfway nto the current board s term. These electons are held mdway between trennal electons (18 months after trennal electon). Another feature of the reforms was the abolton of geographcal school enrolment zones to enable parents to choose whch school ther chldren would attend. Ths created competton between schools to attract the better students. Schools that attracted an overabundance of applcatons were requred to employ ther own enrolment scheme to avod overcrowdng wthn the school. In 2000, an amendment to the Educaton Act 1989 ncluded the requrement that a school s enrolment scheme could not act to exclude any student resdent wthn the zone covered by ts enrolment scheme. 3.3. The School Decle System In 1995, the Mnstry of Educaton ntroduced a system of classfyng schools nto decles based on the soco-economc background of the school s student communtes, to help the Mnstry determne whch schools requred more fundng (Fske and Ladd, 2000). The addtonal fundng asssts schools to overcome barrers to learnng that students n low socoeconomc communtes may face. Decle 1 schools represent the 10% of schools that have the hghest proporton of students from low soco-economc communtes, whle decle 10 schools are the 10% of schools wth the lowest proporton of students drawn from low socoeconomc communtes (Mnstry of Educaton, 2009f). Accordngly, lower decle schools receve larger amounts of fundng. 8
There are fve key factors that are used to measure a school s decle: household ncome 10, occupaton 11, household crowdng 12, educatonal qualfcaton 13, and ncome support 14 (Mnstry of Educaton, 2009f). Informaton collected from the Census s used to calculate these factors for each meshblock 15 that a student of a partcular school resdes n. These fve factors are then weghted by the number of students from each meshblock. Each school s ranked n relaton to all other schools for each of the fve factors and receves a score for the percentle they fall nto. The unweghted total of these fve scores s then used to produce the overall rankng of the school, n relaton to all others n New Zealand. The schools are then allocated nto one of ten decles (Mnstry of Educaton, 2009f). School decles are automatcally recalculated every fve years, followng a new census. However, schools are able to apply for a revew through two crtera. The frst s a change n the physcal catchment area of the school. The second s when a school feels strongly that the soco-economc status of the students wthn the school s catchment area has changed (Mnstry of Educaton, 2009g). 10 Household ncome s the percentage of households wth equvalent ncome (ncome adjusted for the number of adults and chldren n the household and the age of the chldren, but excludes any household that receves a beneft) n the lowest 20% natonally. 11 Occupaton s the percentage of employed parents workng n an occupaton of skll levels 4 or 5 of the Australa and New Zealand Standard Classfcaton of Occupatons (ANZSCO). These generally comprse labourers, machne operators and assemblers, and other lower sklled occupatons rrespectve of the sector nvolved. 12 Household crowdng s the percentage of households wth an equvalsed crowdng ndex greater than one. Ths ndex measures the proporton of household members per bedroom adjusted for the number of chldren under 10 years of age (every two are assumed to share one bedroom). Couples, and others, are each assgned one bedroom. 13 Educatonal qualfcaton measures the percentage of parents wth no tertary or school qualfcatons. 14 Income support s the proporton of parents who drectly receved an ncome beneft (Domestc Purposes, Unemployment, or Sckness and Invald s beneft) n the prevous year, but does not nclude parents who receve Famly Support. 15 A meshblock s the smallest spatal unt used by Statstcs New Zealand. In urban areas t s approxmately the sze of a cty block; n rural areas t s smlar n populaton sze but larger n area. 9
4. Methodology 4.1. Parental Votng Turnout n School BOT Electons We begn our analyss wth a smple weghted 16 OLS regresson of the parental votng rate regressed on the (prvate) homeownershp rate and the state-owned housng rate, 17 wth no addtonal control varables. Ths provdes us wth a smplstc estmate of the relatonshp between prvate (and state) homeownershp and votng partcpaton, as may be referred to by commentators dscussng smple bvarate assocatons between homeownershp and other outcomes. We progressvely ntroduce addtonal control varables to deduce whether the smplstc relatonshps (f they exst) are accounted for by other covarates, rather than homeownershp (or state-owned housng) per se. The frst (smple) specfcaton s presented below: VotngTurnout HO SO (1) where VotngTurnout represents the parental votng partcpaton rate n school s BOT electon; HO s the homeownershp rate n the Census area unt (CAU) 18 that school s located; SO s the state-owned housng rate n the CAU that school s located; β and γ represent the coeffcents on HO and SO respectvely; α s a constant; and ε s the error term. The second specfcaton adds controls for populaton characterstcs n order to ascertan whether the (smple) relatonshps between homeownershp and electon partcpaton hold up after controllng for the nature of the local populaton. Varables ncluded n ths set are: an urban ndcator varable (equal to 1 f school s located n an urban area, zero otherwse); log of CAU populaton per hectare (.e. populaton densty); log of total CAU populaton; proporton of households wth school-aged students; populaton age-band 16 Weghted by school sze. 17 The state, through Housng New Zealand Corporaton, owns approxmately 5% of houses n New Zealand. As shown n Table 4, the state-owned proporton s hgher than 5% for households wth school-aged chldren. 18 A CAU s an aggregaton of meshblocks that, n urban areas, typcally corresponds to a suburb. 10
proportons 19 ; and BOT regon dummes 20 (equal to 1 f school falls n that partcular board regon, zero otherwse).the second OLS specfcaton can be represented as: VotngTurnout HO SO P p 1 POP p p (2) where POP p represents the p th populaton characterstc (lsted above) for school ; and the δ p represent the correspondng coeffcents. School characterstcs are added n the thrd specfcaton as a further robustness check on the homeownershp/votng relatonshp. School characterstcs nclude: decle dummes (equal to 1 f a school s classed n that decle, zero otherwse), school roll numbers, and the number of neghbourng schools wthn the same CAU (a marker of school competton). Equaton 3 represents the thrd specfcaton: VotngTurnout P S p HO SO ppop p 1 s 1 SCHAR s s (3) where SCHAR s represents the s th school characterstc of school from those lsted above; and the φ s represent the correspondng coeffcents. The fnal specfcaton adds controls for local area characterstcs. Local area (CAU) characterstcs nclude: the proporton of resdents new to the communty wthn the last fve years (from wthn New Zealand and overseas); the proporton of mgrants; the proporton of resdents who have partcpated n volunteer work wthn the last four weeks (as a measure of communty socal captal); the dstrbuton of ethnctes; the proportons of structures of households (.e. whether they are couple-parents, sngle-parent or other); proporton of hghest qualfcaton attaned by adults; proporton of ncome benefcares; proporton employed and unemployed; log mean ncome and log standard devaton of ncome for 19 Age-bands were defned as: 17 years or younger, 18 24 years, 25 64 years, and 65 year or older. Age-band 17 years or younger s omtted as the base group. 20 BOT regons nclude: Northland, Auckland, Wakato, Bay of Plenty, Central East, Central West, Wellngton and Wararapa, Marlborough (ncludng Nelson and West Coast), and Southern (combned Canterbury, Otago, and Southland). 11
households wth school-aged chldren; the log mean ncome and log standard devaton of ncome for all households n the area; and the log of the medan captal value of resdental propertes n the CAU. Equaton 4 presents the fourth specfcaton: VotngTurnout P S L p s HO SO ppop sschar p 1 s 1 l 1 LACHAR l l (4) where LACHAR l represents the l th local area characterstc lsted above for school ; and the λ l represent the correspondng coeffcents. 4.2. School BOT Electon To nvestgate the mpact of homeownershp on the probablty that a school wll proceed to a BOT electon, we adopt a smlar methodology to the analyss on the parental partcpaton rate n electons, but we use a weghted probt model nstead of a weghted OLS regresson n estmaton. Ths allows us to obtan the margnal mpacts of homeownershp on the probablty of a school proceedng to a BOT electon. The resultng equatons are smlar to equatons (1) to (4) above, wth replacement of dependent varable VotngTurnout wth Vote (a bnary varable equal to 1 f school held a BOT electon n 2007, or zero otherwse). The four equatons are represented as: P( Vote 1) ( HO SO ) (5) P p p p 1 PVote ( 1) ( HO SO POP ) (6) P S p s p s p 1 s 1 (7) P( Vote 1) ( HO SO POP SCHAR ) P S L p s l p s l p 1 s 1 l 1 (8) P( Vote 1) ( HO SO POP SCHAR LACHAR ) 12
5. Data Descrpton Data were obtaned from two man sources: the Mnstry of Educaton (MoE) and Statstcs New Zealand (SNZ). The school votng and school characterstcs data were sourced from MoE, whle populaton and local area characterstcs were sourced from SNZ. Addtonally, we obtaned the medan captal values of propertes n local areas from Quotable Value New Zealand (QVNZ). The data are explaned n more detal below. 5.1. School Board of Trustees Electon Data MoE took over the collecton and reportng of BOT electons n 2004. It stores records on parents votng paper return rates,.e. the number of votng papers sent out to parents n a BOT electon, along wth the number of votng papers returned. These votng paper return rates represent our dependent varable (parental votng turnout n school BOT electons (VoteTurnout)). MoE were only able to provde data on parent votng paper return rates from the 2007 trennal electons. 21 Many schools were mssng votng results, due to those schools not proceedng to electon. The latter data were used to construct our bnary Vote varable. 5.2. School Characterstcs Data MoE suppled school roll numbers from the July 2007 school roll returns. The data detal the total number of students enrolled n each school. School decle ratngs for 2007 are also provded by MoE, as are data for the attanment level of secondary school leavers for each school. The latter data detal the proporton of students who gan unversty entrance (UE) or better, NCEA level 2 or better, NCEA level 1 or better, and lttle or no formal attanment. We create two attanment varables, hgh qualfcatons and no qualfcatons. Hgh qualfcatons represents the proporton of school leavers wth UE or better qualfcatons; no qualfcatons 21 Votng data are avalable for md-term electons (held n 2005 and 2008), but because not all schools are requred to hold md-term electons, we dd not utlse these md-term results. 13
represents the proporton of students who leave school wth no formal school qualfcaton,.e. those below NCEA level 1. 5.3. Populaton and Local Area Characterstcs Data We use 2006 ndvdual unt record data from SNZ s Census of Populaton and Dwellngs to comple demographc and economc characterstcs of households wth school-aged chldren. 22 Indvdual unt record data s only avalable n the SNZ data laboratory. Gven ts hgh confdentalty, the ndvdual unt record data obtaned from SNZ are subject to SNZ s confdentalty rules. 23 The census varables compled are: an urban/rural dummy, populaton densty, proporton of one-famly households, proporton of people new to a communty, proporton of foregn-born resdents, ethncty proportons, household structure, hghest qualfcatons, proporton of ncome benefcares, employment status, proporton partcpatng n volunteer work, mean dwellng ncome. 5.4. Data Samples for Analyss We analyse four dfferent samples of data, determned by the school type (full prmary, contrbutng prmary, ntermedate, and secondary school). The age brackets of students enrolled dffer across school types. Only parents of students enrolled at a partcular school are able to vote n the BOT electon of that school. Therefore, we restrct each school type s data to measure only households wth students n the approprate age brackets. For each sample, we only consder state and state-ntegrated schools, as these are the only schools requred to hold BOT electons. 22 For a few varables (e.g. populaton densty), the varables pertan to the full CAU populaton. The notes to Tables 3 and 4 detal whch choce of aggregaton s relevant for each varable. 23 These requre that any data are to be randomly rounded to base 3, and small values (less than 3) are omtted. 14
Each school s mapped to the SNZ Census area unt (CAU) that t s located wthn to capture local communty effects. The BOT electon votng data s avalable for 2007, whle the ndvdual unt record Census data s collected n 2006. To account for the one year dfference, we adjust the age brackets of students back one year and select only those households wth approprately aged chldren from the 2006 census of populaton and dwellngs. Table 2 presents the approprate age brackets of chldren n 2006 attendng each school type n 2007. 2006 ndvdual unt record Census data of sutable households are then aggregated to CAU level and matched to schools located wthn that CAU. We then combne the Census data wth the schoolng data to form the sample datasets. 24 6. Results 6.1. Descrptve Statstcs Before presentng estmaton results, we provde descrptve statstcs of all varables. Table 3 presents the proporton of schools that held a BOT electon n 2007, along wth the dstrbuton of the key varables of nterest: electon partcpaton, homeownershp rate, and the proporton of state-owned houses for households wth school-aged chldren n the local area of each school. The weghted mean 25, standard devaton, and percentles of each varable are reported for each sample. Observatons on the partcpaton rate n BOT electons and homeownershp rates are evenly dstrbuted across all samples wth the mean and medan values very smlar for these two varables n each case. The dstrbutons of the proporton of homes that are state-owned are skewed to the left for all samples, wth the mean approxmately equal to the 75 th percentle. 24 In combnng the Census and schoolng data, sx state full prmary schools and three state contrbutng prmary schools were removed due to havng nconsstent census data. The total number of schools used n each sample s provded n Table 2. 25 Means were weghted by school sze (total roll n 2007) so data represent the average pupl. 15
Table 4 presents the weghted means of covarates for each of the four samples. 26 The varables represent weghted mean proportons of the sample populatons, except where geometrc means are used because of the nature of the data, as ndcated n the table. Some nterestng ponts emerge. The parental votng turnout n BOT electons s the hghest for the full prmary school sample, nearly double the rate n ntermedate and secondary schools. Full prmary schools are the most rurally located of the four samples (havng the lowest proporton located n urban areas and populaton per hectare s sgnfcantly lower that other school types) and they have the hghest proporton of state-ntegrated schools. Homeownershp rates are broadly smlar across school type, as s the prevalence of stateowned houses. Proportons of ethnc groups are relatvely constant across the samples, except that fewer Asan households locate n areas near a full prmary school,.e. Asan households are more concentrated n urban areas than are other ethnctes. 6.2. Parental Votng Turnout n School BOT Electons One of our two key objectves s to nvestgate whether a school located n an area of hgh homeownershp rates (for households wth school-aged chldren) experences a hgher rate of parental votng turnout n school BOT electons relatve to ts correspondng school sample. Table 5 reports the weghted OLS regresson results of the relatonshp between homeownershp and state-owned housng varables, and parental votng turnout. Specfcally, we attempt to dentfy whether homeownershp has any effect on parents votng habts. Panel A n Table 5 reports the results from the smple regresson of parental votng turnout on the homeownershp rate and state-owned housng rate wth no addtonal control varables (equaton (1)), to observe whether homeownershp or state-owned housng rates 26 For secondary schools, we separately mapped the data to the local CAU and to the local school zone (only for those schools that had a school zone). We then obtaned both CAU-based estmates and zone-based estmates for VotngTurnout and Vote. Results were smlar, and the much larger sample sze for the CAU-based estmates led us to favour ths approach over the zone-based approach. 16
have a sgnfcant assocaton wth votng rates. Panels B through to D represent results from estmatng equatons (2) (4), where we progressvely ntroduce addtonal control varables. These extensons enable us to nfer whether the smple assocatons are accounted for by other covarates that are themselves correlated wth homeownershp or state-owned housng. From Table 5 Panel A, the homeownershp rate has a postve assocaton wth the parental votng turnout for all schools, and s sgnfcant at the 5% level for full prmary, contrbutng prmary and ntermedate schools. Thus, schools located n areas wth hgher rates of homeownershp are observed to have hgher rates of parental votng turnout n school BOT electons. The effect of state-owned housng s negatve for all but ntermedate schools, and s sgnfcant for full and contrbutng prmary schools. Thus (apart from the ntermedate school sample), schools located n areas wth a hgher prevalence of state-owned housng are observed to have a lower parental votng turnout. The addton of controls for populaton composton, densty and urbansaton n Panel B substantally mproves the explanatory power of the equatons. Smlar effects of homeownershp on parental votng turnout are observed, wth sgnfcant postve effects remanng for full prmary, contrbutng prmary and ntermedate schools. The proporton of state-owned homes now has a postve coeffcent for each sample, but none s sgnfcant. We nclude school characterstcs n Panel C. The magntudes of all homeownershp rate coeffcents are now reduced, wth the secondary school coeffcent now margnally negatve, and all coeffcents lose ther sgnfcance. Smlarly, none of the state-owned housng coeffcents s sgnfcant. Panel D ncludes the full range of covarates. Three of the four homeownershp coeffcents are now negatve and none s sgnfcant. Three of the state-owned housng rate coeffcents are negatve, wth the coeffcent for contrbutng prmary schools beng negatve and sgnfcant at the 5% level. Thus, we observe that a hgher prevalence of state-owned 17
housng n a local area lowers the turnout of parental votng n a contrbutng prmary school BOT vote. The results presented n Table 5 ndcate that whle the rate of homeownershp s postvely assocated wth parental votng turnout n school electons, once controls for populaton, school and local area characterstcs are added, we fnd no mpact of the homeownershp rate on voter turnout. For the state-owned housng rate, we fnd a sgnfcant (negatve) effect only for contrbutng schools once all covarates are added. These results provde a useful cauton regardng use of smple assocatons between homeownershp and outcome varables that may not be robust to controls for other (correlated) varables. The full estmaton results from Panel D are provded n Table 6. For full prmary and secondary schools, we observe that a hgher proporton of dwellngs wth elgble chldren decreases parental votng turnout, and the sze of the school roll s negatvely related to votng turnout. These results are consstent wth a hypothess that parents are less lkely to vote f there are more potental voters present. In these crcumstances, parents may consder that ther own vote wll not nfluence the overall result of the outcome, and they therefore abstan from votng. The decle ndcators mply that votng turnout s postvely related to affluence. Schools wth a larger proporton of students from more (less) soco-economcally deprved communtes receve lower (hgher) rates of votng from parents of those students. The trend s most apparent n prmary schools, although the secondary schools results suggest that hgh decle communtes partcpate more strongly n school electons than low or mddle decle communtes. For secondary schools, a hgh proporton of households that are new to the communty has a sgnfcant negatve effect on the votng rate. Newly resded parents may have a less well-developed sense of the local communty or know fewer canddates, resultng n a lower lkelhood of votng n BOT electons. In prmary schools, sngle-parent 18
households are less lkely to vote, whle a hgher proporton of Maor households reduces voter turnout for full prmary schools. A hgh proporton of people who volunteer wthn the communty s assocated wth a hgher rate of votng n full prmary BOT electons. Ths varable may be proxyng for people wth an nherently hgh level of socal captal and communty nvolvement, wth these same trats carryng through to electon partcpaton. 6.3. Probt Analyss of the Lkelhood of BOT Electons Whle the homeownershp rate s found to have no effect on the parental votng partcpaton rate n school BOT electons once other characterstcs are controlled for, t may affect whether a school proceeds to a BOT electon. The man reason for a school not proceedng to electon s that t receves no more canddates than there are postons avalable, and the school therefore does not requre an electon. 27 From Table 3, approxmately half of each prmary school type and three-quarters of secondary schools proceeded to a BOT electon n 2007. We use probt regressons to nvestgate the possble homeownershp and other nfluences on the probablty that a school proceeded to a BOT electon (as outlned n secton 4). Table 7 presents the results from ths analyss, hghlghtng the margnal effects of the homeownershp rate and state-owned housng rate on the lkelhood that a school wll proceed to a BOT electon. When there are no addtonal covarates (Panel A), we observe that the homeownershp rate s assocated wth an ncreased probablty of a BOT electon beng held for all school types, wth sgnfcant effects for both full prmary and contrbutng prmary schools. The state-owned housng rate ncreases the probablty of a BOT electon for most schools, but only sgnfcantly for full prmary schools. Addng controls for populaton 27 An alternatve (but rare) reason s that a commssoner has been apponted by the Secretary for Educaton to govern the school n place of the BOT, when the BOT has faled to fulfl ts responsbltes. All BOT responsbltes are then transferred to the commssoner. (Educaton Act, 1989). 19
composton, densty and urbansaton (Panel B), we observe smlar results to the prevous model: a sgnfcant postve effect from homeownershp on the probablty of holdng a BOT electon n full prmary and contrbutng prmary schools, and no sgnfcant effect of the state-owned housng rate on the probablty of a BOT electon. Panel C adds controls for school characterstcs. The effect of homeownershp s now no longer sgnfcant for any school type (wth the excepton of a weakly sgnfcant result for contrbutng prmary schools). State-owned housng rates sgnfcantly ncrease the probablty of a BOT electon wthn ntermedate schools (wth a weak effect also for full prmary schools). Panel D adds controls for local area characterstcs. Ths full model, wth all covarates ncluded, ndcates that homeownershp sgnfcantly ncreases the chance that a full prmary or contrbutng prmary school wll hold a BOT electon. Smlarly, a hgh prevalence of state-owned housng sgnfcantly ncreases the chance of a BOT electon for full prmary and ntermedate schools (wth a weak effect also for secondary schools). 28 Thus, we observe that the probablty of a BOT electon beng held n (full and contrbutng) prmary schools s sgnfcantly and postvely affected by homeownershp rates, but the same effect s not observed for schools caterng just to older age groups (ntermedate and secondary schools). The estmated coeffcents mply that full (contrbutng) prmary schools wth a 10 percentage pont hgher local area homeownershp rate are 5.5 (7.9) percentage ponts more lkely to hold a BOT electon. Gven that only around 50% of prmary schools hold a BOT electon these are farly large effect szes. 29 28 We are wary, however, about the estmated sze of the margnal effects relatng to the ntermedate school sample snce that sample s small (N=121) and we have 40 explanatory varables n the equaton. 29 These results suggest that, f anythng, our pror estmates for the mpact of homeownershp on BOT electon turnout (Table 5) could be based upwards as we may expect homeownershp to have a larger mpact on BOT electon turnout n schools where hgh homeownershp has ncreased the lkelhood of havng electons n the frst place. These fndngs therefore renforce our pror concluson that homeownershp, per se, has no mpact on voter turnout n a BOT electon. 20
One explanaton for these results may be that homeownershp ncreases people s sense of communty, snce t has been shown to ncrease ther enthusasm towards beng nvolved n local communty affars (Roskruge et al, 2011). A hgher rate of homeownershp may therefore lead to more parents standng for the board, ncreasng the probablty of a school holdng an electon. The spatal catchment of prmary schools s generally more focused on a local communty than are those for ntermedate and secondary schools. Homeowners may be less enthusastc n standng for these larger schools that servce multple communtes, as they do not feel as great an affnty or afflaton wth those outsde ther mmedate local communty. The sgnfcant (postve) coeffcent on the state-owned housng rate accords wth the fndng n Roskruge et al (2011) that state tenants have a statstcally smlar sense of communty to homeowners, wth each havng a sgnfcantly greater sense of communty than prvate renters. The estmated coeffcents mply that full prmary schools wth a 1 percentage pont hgher local area state-owned housng rate are 1 percentage pont more lkely to hold a BOT electon. The mean school only has 6 percent of the local housng beng state-owned so effect szes are smlar to those for homeownershp. These results suggest that the securty of the long-term tenances avalable to state tenants (relatve to prvate tenants) has postve socal captal spn-offs for the local communty. The full set of results of the probt regresson wth all covarates ncluded s presented n Table 8. The sze of a school has a sgnfcant and postve effect on the probablty of a school holdng an electon; ths apples to all school types analysed. Larger schools wll have a greater pool of parents from whch potental canddates for the board may arse and, therefore, less chance that the school wll have a lack of canddates to force an electon. A consstent fndng wth the VoteTurnout results s that a school s more (less) lkely to hold an electon f t s hgh (low) decle. Hgh (low) soco-economc parents may have more (less) 21
self-esteem, or avalable tme, than other parents and therefore be more wllng to put themselves forward as a canddate. 7. Conclusons Many countres have mplemented polces to boost homeownershp rates wth the am of mprovng household and communty outcomes. Pror evdence ndcates that homeowners are more poltcally actve than renters and have hgher votng rates n poltcal electons. Homeowners are also less moble and have nvested a large fnancal stake n ther own property, and are therefore more ncentvsed to mprove the qualty of ther neghbourhood. We have analysed one aspect of poltcal partcpaton and cvc effort, nvestgatng the effect of homeownershp rates on parental votng turnout and parental partcpaton n school board of trustees electons. All state and state-ntegrated schools n New Zealand are requred to hold a board of trustees electon trennally, and parents are able to stand for, and elect, parental representatves to the board. Parental votng s not compulsory, and schools receve wdely varyng rates of parental votng. We utlse data on the 2007 school board of trustees electons to estmate the effect of homeownershp on parental votng turnout n schools. Four samples of data were analysed, one for each of the school types: full prmary, contrbutng prmary, ntermedate, and secondary school. Smple weghted OLS regressons, where only the homeownershp rate and the rate of state-owned housng were controlled for, found a postve assocaton between homeownershp and parental votng turnout n all school types. There was a negatve assocaton between the state-owned housng rate and the parental votng turnout. The effects on voter turnout of homeownershp and of state-owned housng rates fall away once addtonal controls are ncluded for populaton, school and local area characterstcs. The mplcatons of these estmates are that homeownershp has no dscernble effect on the parental votng turnout once other factors, such as school sze and decle ratngs are 22
controlled for. State-ownershp rates are found to have a negatve mpact on votng turnout only for contrbutng prmary schools once other factors are controlled for. Homeownershp, however, consstently affects the chance that a prmary school proceeds to a school BOT electon. Based on pror lterature, we conjecture that homeownershp ncreases owners sense of communty and, therefore, ncreases ther wllngness to stand as a canddate for the board. However, ths behavour does not carry through to ntermedate and secondary schools, whch generally servce larger communtes. As the communty sze ncreases, the affnty and enthusasm for homeowners to stand for a board decreases, hence decreasng the probablty that these schools proceed to a BOT electon. As the state-housng rate ncreases, the probablty of proceedng to an electon agan rses (for three of the four school types). The results for the probablty of holdng an electon are consstent wth the theory that housng stablty (ether through owner-occupaton or through a state tenancy) leads to a greater sense of communty and cvc nvolvement for those n such forms of tenure. However, ths effect seems lmted to drect personal partcpaton n school affars. We fnd no broader benefts of homeownershp or of state-ownershp n enhancng parents nvolvement n partcpatory democracy n the form of school voce. 23
8. References Boehm, Thomas P. and Alan M. Schlottmann. 1999. Does Home Ownershp by Parents Have an Economc Impact on Ther Chldren? Journal of Housng Economcs, 8, pp. 217-232. Detz, Robert D. and Donald R. Haurn. 2003. The socal and prvate mcro-level consequences of homeownershp, Journal of Urban Economcs, 54, pp 401-450. Dpasquale, Dense and Edward L. Glaeser. 1999. Incentves and Socal Captal: Are Homeowners Better Ctzens? Journal of Urban Economcs, 45, pp. 345-384. Educaton Act, 1989. Educaton Act 1989. Avalable onlne at http://www.legslaton.govt.nz/act/publc/1989/0080/latest/dlm175959.html?search= ts_act_educaton_resel&p=1&sr=1. Last accessed 11 December 2009. Engelhardt, Gary V., Mchael D. Erksen, Wllam G. Gale, and Gregory B. Mlls. 2010. What are the socal benefts of homeownershp? Expermental evdence for lowncome households, Journal of Urban Economcs, 67, pp. 249-258. Fske, Edward. and Helen Ladd, 2000. When Schools Compete: A Cautonary Tale, 1 st ed., Washngton D.C.: Brookngs Insttutonal Press. Green, Rchard K. and Mchelle J. Whte. 1997. Measurng the Benefts of Homeownng: Effects on Chldren, Journal of Urban Economcs, 41, pp. 441-461. Hanfan, L.J.. 1916. The Rural School Communty Center, Annals of the Amercan Academy of Poltcal and Socal Scences, 67, pp. 130-138. Haurn, Donald R., Toby L. Parcel and R. Jean Haurn. 2002. Does Homeownershp Affect Chld Outcomes? Real Estate Economcs, 30:4, pp. 635-666. Koff, Karla and Arjt Sen. 2005. Homeownershp, Communty Interactons, and Segregaton, The Amercan Economc Revew, 95:4, pp. 1167-1189. 24
Manturuk, Km, Mark Lndblad, and Roberto G. Querca. 2009. Homeownershp and Local Votng n Dsadvantaged Urban Neghbourhoods, Ctyscape: A Journal of Polcy Development and Research, 11:3, pp. 213-230. Mnstry of Educaton, 2009a. Types of schools. Avalable onlne at http://www.mnedu.govt.nz/parents/allages/educatoninnz/schoolsinnewzealand/s chooltypes.aspx. Last accessed 11 December 2009. Mnstry of Educaton, 2009b. Schoolng Structures and Governance Optons. Avalable onlne at http://www.mnedu.govt.nz/boards/effectvegovernance/flexbltyboard Structures/SchoolngStructuresAndGovernanceOptons.aspx. Last accessed 11 December 2009. Mnstry of Educaton, 2009c. Consttutons for boards of trustees. Avalable onlne at http://www.mnedu.govt.nz/boards/effectvegovernance/flexbltyboardstructures/ ConsttutonsForBOTs.aspx. Last accessed 11 December 2009. Mnstry of Educaton, 2009d. Overvew. Avalable onlne at http://www.mnedu. govt.nz/boards/effectvegovernance/overvew.aspx. Last accessed 11 December 2009. Mnstry of Educaton, 2009e. School Board of Trustees Electons, Co-optons and Selecton to a casual vacancy. Avalable onlne at http://www.mnedu.govt.nz/ Boards/EffectveGovernance/BoardElectons.aspx. Last accessed 11 December 2009. Mnstry of Educaton, 2009f. How the decle s calculated. Avalable onlne at http://www.mnedu.govt.nz/nzeducaton/educatonpolces/schools/schooloperato ns/resourcng/operatonalfundng/decles/howthedecleiscalculated.aspx. Last accessed 11 December 2009. Mnstry of Educaton, 2009g. Revew of Decles General Informaton. Avalable onlne at http://www.mnedu.govt.nz/nzeducaton/educatonpolces/schools/schooloperato 25
ns/resourcng/operatonalfundng/decles/revewofdeclesgeneralinformaton.asp x. Last accessed 11 December 2009. Roskruge, Matthew, Arthur Grmes, Phlp McCann & Jacques Poot. 2011. Homeownershp and Socal Captal n New Zealand, Motu Workng Paper 11-02, Wellngton: Motu. Ross, Peter H. and Eleanor Weber. 1996. The Socal Benefts of Homeownershp: Emprcal Evdence from natonal Surveys, Housng Polcy Debate, 7:1, pp. 1-35. Tax Workng Group. 2010. A Tax System for New Zealand s Future, Centre for Accountng, Governance and Taxaton Research, Wellngton: Vctora Unversty of Wellngton. 26
Table 1: 2007 School Numbers and Rolls by Governance and School Type Full Prmary Contrbutng Prmary Intermedate Secondary Other TOTAL State State-Integrated Other TOTAL Schools 888 744 121 194 203 2,150 Pupls 136,901 199,495 57,087 194,834 56,149 644,466 Schools 184 49 0 28 65 326 Pupls 29,458 9,339 0 11,507 34,158 84,462 Schools 29 0 2 12 64 107 Pupls 5,380 0 192 4,403 20,802 30,777 Schools 1,101 793 123 234 332 2,583 Pupls 171,739 208,834 57,279 210,744 111,109 759,705 Proporton n the Analyss Sample * Schools 0.97 1.00 0.98 0.95 NA 0.85 Pupls 0.97 1.00 1.00 0.98 NA 0.84 Note: Sourced from the Mnstry of Educaton. The box nsde the bold lnes consttutes the analyss sample for the remander of the paper. * The analyss samples n our study only consder state and state-ntegrated schools across the school types shown. The proportons do not sum perfectly for full prmary and contrbutng prmary schools, due to the loss of sx state full prmary schools and three contrbutng prmary schools n the formaton of our analyss samples.
Table 2: Data Samples used n Analyss Full Prmary Contrbutng Prmary Intermedate Secondary Number of Schools 1,066 790 121 222 Years of Students Year 1 Year 8 Year 1 Year 6 Year 7 Year 8 Year 9 Year 13 Age of chldren n households (2006) 4 11 years 4 9 years 10 11 years 12 16 years Aggregaton Level Census Area Unt Census Area Unt Census Area Unt Census Area Unt
Table 3: The Dstrbuton of Electon Partcpaton and Homeownershp across Students Full Prmary Contrbutng Prmary Intermedate Secondary Proporton of Schools wth a Board of Trustee Electon Proporton 0.48 0.55 0.50 0.75 Partcpaton Rate n Board of Trustee Electons for each School wth an Electon Mean 0.38 0.31 0.22 0.22 Std. Dev. 0.16 0.12 0.13 0.14 10th Percentle 0.20 0.17 0.13 0.12 25th Percentle 0.27 0.23 0.15 0.15 50th Percentle 0.36 0.30 0.18 0.19 75th Percentle 0.48 0.37 0.23 0.26 90th Percentle 0.60 0.46 0.28 0.30 Homeownershp Rate n the Local Area of each School Mean 0.60 0.58 0.56 0.60 Std. Dev. 0.16 0.16 0.15 0.14 10th Percentle 0.38 0.36 0.34 0.41 25th Percentle 0.50 0.47 0.47 0.52 50th Percentle 0.63 0.59 0.58 0.62 75th Percentle 0.72 0.70 0.67 0.69 90th Percentle 0.80 0.77 0.75 0.76 State-ownershp Rate n the Local Area of each School Mean 0.06 0.07 0.09 0.07 Std. Dev. 0.10 0.11 0.11 0.10 10th Percentle 0.00 0.00 0.00 0.00 25th Percentle 0.00 0.00 0.00 0.01 50th Percentle 0.02 0.03 0.05 0.04 75th Percentle 0.08 0.09 0.11 0.09 90th Percentle 0.17 0.22 0.26 0.19 Number of Schools 1,066 790 121 222 Note: All results except the frst panel are weghted by school sze so that the data represent the dstrbuton across pupls. Homeownershp and state-owned housng rates are calculated for households wth school-aged chldren n the area unt n whch the school s located.
Table 4: Mean Characterstcs of Students across Schools Full Prmary Contrbutng Prmary Intermedate Secondary BoT Electon Partcpaton Rate 0.38 0.31 0.22 0.22 State-Integrated School 0.18 0.04 0.00 0.06 School Had a Md-Term Electon 0.06 0.07 0.03 0.02 Mean School Roll (geometrc) 228 327 521 1,099 School Leavers wth No Qualfcatons 0.18 School Leavers wth Hgh Qualfcatons 0.26 In an Urban Area 0.78 0.95 1.00 0.99 Mean Populaton (geometrc) 2,623 3,245 3,376 3,191 Mean Pop per Hectare (geometrc) 3.19 11.83 19.49 14.05 Populaton Age <18 0.31 0.31 0.30 0.30 Populaton Age 18-24 0.09 0.10 0.12 0.12 Populaton Age 25-64 0.59 0.58 0.57 0.57 Populaton Age 65+ 0.02 0.02 0.01 0.01 Dwellngs wth Elgble Chldren 0.23 0.18 0.08 0.17 Homeownershp Rate 0.60 0.58 0.56 0.60 State-Owned Housng Rate 0.06 0.07 0.09 0.07 New to Communty n Last 5 Years 0.60 0.65 0.59 0.56 Foregn-born 0.20 0.25 0.29 0.29 Volunteered n Last Four Weeks 0.19 0.17 0.16 0.14 Pakeha Ethncty 0.52 0.47 0.41 0.47 Maor Ethncty 0.21 0.19 0.20 0.18 Pacfc Island Ethncty 0.07 0.10 0.11 0.09 Asan Ethncty 0.06 0.11 0.14 0.14 Other Ethncty 0.14 0.13 0.13 0.13 Couple wth Chldren Household 0.69 0.68 0.64 0.62 Sngle Parent Households 0.22 0.23 0.27 0.28 Other Households 0.08 0.09 0.09 0.09 No Qualfcatons 0.22 0.19 0.22 0.24 School Qualfcatons 0.33 0.33 0.34 0.37 Post-school Qualfcatons 0.23 0.23 0.21 0.18 Unversty Qualfcatons 0.13 0.16 0.14 0.12 Employment Rate 0.70 0.68 0.67 0.65 Unemployment Rate 0.04 0.04 0.05 0.06 Recevng Socal Benefts 0.14 0.14 0.14 0.11 Mean Dwellng Income (geometrc) 45,694 45,783 44,330 46,089 Std Dev Log Dwellng Income 0.68 0.70 0.72 0.74 Mean Dwellng Income n Area (geometrc) 40,770 40,891 38,662 39,034 Std Dev Log Dwellng Inc n Area 0.73 0.75 0.77 0.77 Medan House Value n Area 255,304 309,976 307,804 306,630 Sample Sze 1,066 790 121 222 Note: All results are weghted by school sze so that the data represent the average pupl. All varables are calculated for the area unt n whch the school s located. All characterstcs except school characterstcs, populaton (ncludng components), area ncomes and house values are defned over the sample of adults wth elgble schoolaged chldren.
Table 5: Weghted OLS Regresson of BoT Partcpaton Rates versus Homeownershp Full Prmary Contrbutng Prmary Intermedate Secondary Panel A: No Addtonal Control Varables Homeownershp Rate 0.141* 0.215** 0.392* 0.072 (0.061) (0.056) (0.164) (0.104) State-owned Housng Rate -0.311* -0.182* 0.120-0.131 (0.111) (0.081) (0.127) (0.149) R-Squared 0.142 0.164 0.148 0.040 Panel B: Addng Controls for Populaton Composton, Densty and Urbansaton to Panel A Homeownershp Rate 0.155* 0.200** 0.273+ 0.065 (0.057) (0.058) (0.150) (0.140) State-owned Housng Rate 0.126 0.037 0.061 0.006 (0.095) (0.075) (0.165) (0.164) R-Squared 0.545 0.395 0.382 0.184 Panel C: Addng Controls for School Characterstcs to Panel B Homeownershp Rate 0.057 0.052 0.250-0.077 (0.052) (0.060) (0.203) (0.176) State-owned Housng Rate 0.031-0.022 0.306 0.013 (0.104) (0.076) (0.279) (0.154) R-Squared 0.660 0.520 0.526 0.299 Panel D: Addng Controls for Local Area Characterstcs to Panel C Homeownershp Rate -0.003-0.016 0.061-0.643 (0.067) (0.097) (0.540) (0.401) State-owned Housng Rate -0.042-0.175* 1.091-0.298 (0.103) (0.087) (0.665) (0.234) R-Squared 0.694 0.559 0.825 0.401 Number of Schools 512 437 61 167 Notes: ** ndcates sgnfcance at 1% level of sgnfcance. * ndcates sgnfcance at the 5% level of sgnfcance. + ndcates sgnfcance at the 10% level of sgnfcance. Numbers contaned wthn parentheses represent standard errors. School characterstcs added n Panel C nclude: decle dummy varables (decle 5 omtted as base) and the log of school roll. Local area characterstcs added n Panel D nclude: one-famly households, new to communty wthn last fve years, foregn-born, ethncty categores (Pakeha omtted as base), household structure categores (couple households omtted as base), hghest qualfcaton categores (no qualfcatons omtted as base), ncome benefcares, employment statuses, volunteer work wthn the last four weeks, and log of mean dwellng ncome.
Table 6: Weghted OLS Regresson of BoT Partcpaton Rates Full Results Full Prmary Contrbutng Prmary Intermedate Secondary Coef S.E. Coef S.E. Coef S.E. Coef S.E. Homeownershp Rate -0.003 (0.067) -0.016 (0.097) 0.061 (0.540) -0.643 (0.401) State-owned Housng Rate -0.042 (0.103) -0.175* (0.087) 1.091 (0.665) -0.298 (0.234) Urban/Rural -0.028 (0.020) -0.040 (0.027) 0.105 (0.072) Log Populaton -0.008 (0.008) 0.025* (0.013) -0.069 (0.122) -0.033 (0.029) Log Populaton per Hectare 0.002 (0.005) -0.009 (0.005) -0.010 (0.080) -0.003 (0.024) Pop Age 18-24 (vs < 18) -0.306 (0.213) -0.056 (0.149) 0.371 (1.99) -1.57* (0.689) Pop Age 25-64 (vs < 18) 0.020 (0.169) -0.028 (0.213) 1.375 (2.09) -1.65* (0.497) Pop 65+ (vs < 18) -1.83 (1.522) 1.094 (1.46) 28.06 (20.7) -9.61* (3.414) Has Elgble Chldren -0.188 + (0.102) 0.157 (0.202) -1.653 (3.514) -1.84* (0.698) State-Integrated School 0.096** (0.018) 0.034 (0.025) -0.020 (0.052) Had Md-Term Electon -0.003 (0.016) -0.006 (0.016) 0.028 (0.046) Log School Roll -0.117** (0.010) -0.104** (0.013) 0.044 (0.105) -0.107* (0.043) Boardng Facltes -0.043 (0.027) # Neghbourng Schools -0.002 (0.002) -0.003** (0.001) -0.027 (0.024) 0.001 (0.006) Leavers w/ No Quals -0.015 (0.128) Leavers w/ Hgh Quals 0.085 (0.120) Decle 1 School (vs 5) -0.072* (0.026) -0.105** (0.025) -0.180 (0.257) -0.076 (0.130) Decle 2 School (vs 5) -0.027 (0.024) -0.064* (0.020) -0.133 (0.228) -0.026 (0.070) Decle 3 School (vs 5)) 0.000 (0.024) -0.049* (0.024) -0.042 (0.159) -0.009 (0.052) Decle 4 School (vs 5) -0.011 (0.021) -0.033* (0.017) -0.111 (0.137) 0.060 (0.054) Decle 6 School (vs 5) 0.034 (0.021) 0.012 (0.020) -0.066 (0.170) 0.078 (0.051) Decle 7 School (vs 5) 0.042* (0.019) 0.051 + (0.026) 0.040 (0.211) 0.076 (0.056) Decle 8 School (vs 5) 0.044* (0.022) 0.037 + (0.019) 0.225 (0.182) 0.130* (0.065) Decle 9 School (vs 5) 0.057* (0.021) 0.042* (0.020) -0.177 (0.239) 0.220* (0.092) Decle 10 School (vs 5) 0.080** (0.023) 0.064* (0.023) -0.007 (0.221) 0.187* (0.090) New to Communty 0.013 (0.088) 0.038 (0.101) -0.831 (0.560) -0.512 + (0.287) Foregn-born -0.109 (0.177) -0.096 (0.125) -1.083 (0.777) -0.649 (0.437) Volunteered n Last 4 Weeks 0.440* (0.154) 0.066 (0.156) -3.011 (2.226) 0.287 (0.627) Maor (vs Pakeha) -0.154 + (0.085) 0.000 (0.075) -0.519 (0.695) -0.226 (0.320) Pacfc Island (vs Pakeha) 0.032 (0.132) 0.031 (0.094) -0.411 (0.756) 0.238 (0.425) Asan (vs Pakeha) -0.082 (0.122) -0.152 (0.098) 0.370 (0.692) 0.492 (0.357) Other (vs Pakeha) -0.075 (0.178) 0.125 (0.215) -1.658 (1.066) -0.103 (0.647) Sngle-parent (vs Couples) -0.369* (0.129) -0.359* (0.156) -0.073 (0.746) -0.274 (0.279) Other-parent (vs Couples) 0.065 (0.178) -0.061 (0.165) -0.510 (1.539) 0.076 (0.619) School Quals (vs No Quals) -0.054 (0.180) -0.152 (0.169) 0.990 (1.075) 0.693 (0.626) Post-school (vs No Quals) -0.441* (0.171) -0.426* (0.203) 1.072 (1.414) 0.015 (0.788) Unversty (vs No Quals) -0.048 (0.185) -0.005 (0.155) 1.572 (1.196) 0.217 (0.684) Mssng Quals (vs No Quals) -0.034 (0.250) -0.467 + (0.274) 2.987 (1.944) 0.145 (0.909) Recevng Socal Benefts 0.077 (0.231) 0.086 (0.206) -0.833 (1.298) -0.783 (0.650) Employed -0.024 (0.199) 0.013 (0.150) -1.053 (1.390) -0.852 (0.677) Unemployed 0.204 (0.396) -0.024 (0.330) 0.170 (2.105) 0.329 (1.576) Mean Log Dwellng Income -0.136 (0.099) -0.029 (0.137) 0.646 (0.578) -0.096 (0.243) S.D. Log Dwellng Income -0.040 (0.081) -0.001 (0.083) 0.125 (0.402) -0.228 (0.225) Mean Log Dwell Inc (Pop) 0.067 (0.077) -0.162 (0.109) -0.259 (0.360) 0.062 (0.196) S.D. Log Dwell Inc (Pop) 0.189 + (0.113) 0.097 (0.126) 0.827 (0.776) -0.259 (0.353) Log Medan House Value 0.006 (0.015) 0.032 (0.021) 0.002 (0.174) -0.040 (0.058) R-Squared 0.694 0.559 0.825 0.401 Sample Sze 512 437 61 167 Notes: + sgnfcant at 10%; * sgnfcant at 5%; ** sgnfcant at 1%. Robust standard errors are n parentheses and all estmates are weghted by school sze and hence represent the relatonshps for the average pupl. Board regons were control for, but are not presented.
Table 7: Margnal Effects of Homeownershp on the Probablty of BOT Electon Full Prmary Contrbutng Prmary Panel A: No Addtonal Control Varables Intermedate Secondary Homeownershp Rate 0.521* 0.447* 0.299 0.133 (0.170) (0.180) (0.508) (0.284) State-owned Housng Rate 0.488* -0.020 0.630 0.071 (0.289) (0.261) (0.647) (0.372) Panel B: Addng Controls for Populaton Composton, Densty and Urbansaton to Panel A Homeownershp Rate 0.497* 0.752** 0.408 0.438 (0.193) (0.208) (0.569) (0.308) State-owned Housng Rate 0.479 0.110 0.875 0.472 (0.332) (0.308) (0.764) (0.416) Panel C: Addng Controls for School Characterstcs to Panel B Homeownershp Rate 0.162 0.439 + 0.163 0.350 (0.217) (0.233) (0.698) (0.279) State-owned Housng Rate 0.681 + 0.549 1.959* 0.430 (0.360) (0.343) (0.893) (0.340) Panel D: Addng Controls for Local Area Characterstcs to Panel C Homeownershp Rate 0.546* 0.791* -0.234-0.031 (0.262) (0.340) (1.449) (0.296) State-owned Housng Rate 0.992* 0.277 3.159* 0.494 + (0.354) (0.426) (1.566) (0.288) Number of Schools 1,066 790 121 222 Notes: ** ndcates sgnfcance at 1% level of sgnfcance. * ndcates sgnfcance at the 5% level of sgnfcance. + ndcates sgnfcance at the 10% level of sgnfcance. Numbers contaned wthn parentheses represent robust standard errors. School characterstcs added n Panel C nclude: decle dummy varables (decle 5 omtted as base) and the log of school roll. Local area characterstcs added n Panel D nclude: one-famly households, new to communty wthn last fve years, foregn-born, ethncty categores (Pakeha omtted as base), household structure categores (couple households omtted as base), hghest qualfcaton categores (no qualfcatons omtted as base), ncome benefcares, employment statuses, volunteer work wthn the last four weeks, and log of mean dwellng ncome.
Table 8: Margnal Effects on the Probablty of BOT Electon Full Results Full Prmary Contrbutng Prmary Intermedate Secondary Coef S.E. Coef S.E. Coef S.E. Coef S.E. Homeownershp Rate 0.546* (0.262) 0.791* (0.340) -0.234 (1.449) -0.031 (0.296) State-owned Housng Rate 0.992* (0.354) 0.277 (0.426) 3.159* (1.566) 0.494 + (0.288) Urban/Rural -0.077 (0.074) -0.139 (0.115) -0.037 (0.109) Log Populaton -0.036 (0.031) -0.032 (0.046) 0.002 (0.213) 0.065 + (0.035) Log Populaton per Hectare -0.025 (0.019) 0.005 (0.025) 0.350 (0.233) -0.041* (0.017) Pop Age 18-24 (vs < 18) -0.200 (0.793) 1.610* (0.739) 1.875 (3.622) 0.943 + (0.558) Pop Age 25-64 (vs < 18) 0.475 (0.712) 1.188 (0.877) 3.620 (4.865) 1.368* (0.604) Pop 65+ (vs < 18) -6.372 (5.579) 3.429 (7.123) 74.40* (34.72) 3.274 (3.639) Has Elgble Chldren 0.438 (0.514) 0.661 (0.969) 6.945 (7.736) 0.690 (0.610) State-Integrated School -0.016 (0.064) -0.125 (0.087) 0.042 (0.049) Had Md-Term Electon -0.019 (0.090) -0.133 (0.087) -0.029 (0.095) Log School Roll 0.236** (0.036) 0.277** (0.048) 0.433 + (0.255) 0.171** (0.043) Boardng Facltes 0.039 (0.035) Neghbourng Schools 0.006 (0.008) -0.009 (0.006) 0.107* (0.044) 0.009 + (0.005) Leavers w/ No Quals -0.050 (0.149) Leavers w/ Hgh Quals -0.104 (0.168) Decle 1 School (vs 5) -0.096 (0.117) -0.194 (0.121) -2.076* (0.634) -0.006 (0.091) Decle 2 School (vs 5) -0.282* (0.105) -0.182 + (0.100) -1.113* (0.490) 0.013 (0.055) Decle 3 School (vs 5)) -0.194 + (0.106) -0.067 (0.090) 0.516 + (0.303) 0.013 (0.058) Decle 4 School (vs 5) -0.189* (0.092) -0.124 (0.093) -0.108 (0.277) -0.058 (0.046) Decle 6 School (vs 5) 0.058 (0.094) 0.021 (0.098) 0.479 (0.303) 0.078 (0.051) Decle 7 School (vs 5) -0.074 (0.098) 0.066 (0.091) 1.242* (0.438) 0.019 (0.051) Decle 8 School (vs 5) -0.058 (0.099) 0.048 (0.106) 1.882** (0.476) 0.004 (0.052) Decle 9 School (vs 5) -0.048 (0.098) -0.007 (0.104) 1.455* (0.482) -0.070 (0.059) Decle 10 School (vs 5) 0.107 (0.106) 0.134 (0.116) 2.843** (0.698) 0.184* (0.079) New to Communty 0.218 (0.363) 0.091 (0.393) 0.969 (1.274) -0.204 (0.270) Foregn-born -0.278 (0.745) -0.157 (0.629) -6.241* (2.222) 0.488 (0.427) Volunteered n Last 4 Weeks 0.362 (0.549) -0.327 (0.698) -0.014 (2.949) 0.590 (0.598) Maor (vs Pakeha) -0.553 + (0.312) -0.034 (0.369) -0.035 (1.323) -0.503 + (0.268) Pacfc Island (vs Pakeha) -0.958 + (0.528) 0.102 (0.503) 5.303* (1.918) -0.584 + (0.311) Asan (vs Pakeha) 0.313 (0.662) 0.160 (0.480) 9.685** (2.937) -0.333 (0.377) Other (vs Pakeha) 0.010 (0.649) 1.205 + (0.636) 7.442* (2.900) -0.779 (0.549) Sngle-parent (vs Couples) 0.887 + (0.528) -1.271* (0.628) -4.679* (1.492) 0.680 + (0.386) Other-parent (vs Couples) 1.042 (0.732) -0.525 (0.848) 5.367* (2.428) 0.989 (0.661) School Quals (vs No Quals) 0.328 (0.723) -0.615 (0.733) -2.951 (2.550) -0.667 (0.566) Post-school (vs No Quals) -1.082 (0.738) -1.019 (0.802) -10.53* (3.315) 0.543 (0.649) Unversty (vs No Quals) -0.514 (0.679) -0.400 (0.662) -6.049* (2.825) 0.307 (0.477) Mssng Quals (vs No Quals) -0.500 (1.036) 0.143 (1.044) -8.299* (3.404) 0.093 (0.842) Recevng Socal Benefts -0.475 (0.780) 1.833* (0.912) 2.877 (2.992) 0.262 (0.748) Employed -0.277 (0.700) 0.543 (0.736) 9.874** (2.315) 0.305 (0.550) Unemployed -0.012 (1.514) -0.594 (1.487) 11.66* (4.995) -0.972 (1.192) Mean Log Dwellng Income -0.538 (0.395) -0.103 (0.454) -0.054 (1.176) 0.589* (0.270) S.D. Log Dwellng Income -0.553 + (0.310) 0.600 + (0.331) 0.423 (0.822) 0.245 (0.242) Mean Log Dwell Inc (Pop) 0.132 (0.330) -0.313 (0.366) -1.594 (1.391) -0.522* (0.230) S.D. Log Dwell Inc (Pop) 0.440 (0.573) -0.245 (0.656) -3.163 (2.757) -0.593 (0.411) Log Medan House Value 0.085 (0.060) 0.113 (0.079) -0.029 (0.358) -0.123* (0.053) Sample Sze 1,066 790 121 222 Notes: * sgnfcant at 10%; ** sgnfcant at 5%; *** sgnfcant at 1%. Robust standard errors are n parentheses and all estmates are weghted by school sze and hence represent the relatonshps for the average pupl. Board regons were control for, but are not presented.