Bruce D. Baker, AERA 2009 San Diego 1 Does Title I Make the Rich Richer? Bruce D. Baker Rutgers...
-
Upload
gwendolyn-craig -
Category
Documents
-
view
218 -
download
1
Transcript of Bruce D. Baker, AERA 2009 San Diego 1 Does Title I Make the Rich Richer? Bruce D. Baker Rutgers...
1
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Does Title I Make the Rich Richer?
Bruce D. Baker
Rutgers University
2
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Summarizing the Critique(critique of the critique)
• Title I funding varies widely and irrationally across states.– Small states get too much, despite relatively low poverty, because of
minimums• (yes, a problematic political payoff. But redistributing T1 aid to Vermont
and Wyoming doesn’t get you very far)– Rich states get too much, because of the state spending factor
• (But the state spending factor is largely offset by regional cost variation)– In general, richer states and wealthier suburban and urban centers
gain at the expense of poorer (higher poverty) states and rural districts.
• (Equating rural and urban poverty and the effects on educational outcomes is problematic. So too is equating a given poverty rate in NY state and in TX).
• On top of all of this, states and local districts fail to distribute resources equitably to schools.
• (Yes, this needs to be fixed, but it is only one piece of the larger puzzle, and many districts are unable to reshuffle their own resources given their relative position in the state system. Fixing state systems comes first!)
3
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
High Poverty, Rural Black-Belt South
High Poverty, Hispanic Immigrant, Rural & Urban
Southwest
Mixed Poverty, Urbanized Northeast & Great Lakes
Poverty Distribution across the US
4
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Consistently High Competitive Wage/Cost of
Living
Rural/Urban Variation in Competitive Wage/Cost of
Living
Competitive Wage Variation across the US
5
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Original ratio = 2.073CWI Adj. ratio = 1.522
Fayetteville, NC = 1.1408
6773/1.1408 = 5937
Alexandria, VA = 1.5539
14040/1.5539 = 9035
A Second Look at Cameron and PonderosaAnd the Tragic Flaw
6
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
High Poverty Rural Areas High Poverty Urban Core
Within State Poverty Distribution
7
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Low Cost (Wage/COL) Rural Areas
High Cost (Wage/COL)Urban Core
Within State Competitive Wage Distribution
8
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Alabama
Alaska
ArizonaArkansas
California
ColoradoConnecticut
Delaware
Florida
Idaho
Illinois
IndianaIowa
Kansas
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New HampshireNew Jersey
New Mexico
New York
North Dakota
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South Dakota
Tennessee
TexasUtah
Vermont
Virginia
Washington
West Virginia
WisconsinWyoming
260
270
280
290
NA
EP
Mat
h &
Rea
ding
200
7
.1 .15 .2 .25 .3Census Poverty Rate
NAEP Reading&Math 2007 Fitted values
State poverty “effect” on NAEP
The Goal of Poverty Based Funding is to Offset Poverty Effects on Outcomes
9
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
-0.35
-0.30
-0.25
-0.20
-0.15
-0.10
-0.05
0.00
Large City Midsize CityFringe of
Large CityFringe of
Midsize City Large Town Small TownRural Outside
CBSARural Inside
CBSA
NCES Locale Code
Eff
ect
(Coe
f.)
of S
ub
. L
un
ch R
ate
on O
utc
omes
(3
to 5
)
Analysis includes elementary school campus level data 2003-2007 grade 3 through 5 proficiency rates with subsidized lunch estimates conditional on year, limited English proficient concentration and special education rate
The relationship between poverty and outcomes varies by setting!
At the same poverty rate, outcomes are lower in large
cities than in many other settings!
11
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
100
200
300
400
Titl
e I p
er P
upil
.1 .15 .2 .25 .3Census Poverty Rate
Title I funding per pupil positively associated with state poverty rate, but not as systematic as one might expect
Title I per Pupil and Census Poverty
12
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
500
100
015
00
200
025
00
300
0T
itle
I per
Pov
ert
y P
upil
.1 .15 .2 .25 .3Census Poverty Rate
Small State Effect
Relatively Wide Range, Negative Association with Poverty
Title I per Poverty Pupil and Census Poverty
13
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
500
1000
1500
2000
2500
3000
Adj
. Titl
e I p
er P
over
ty P
upil
.1 .15 .2 .25 .3Census Poverty Rate
Small State Effect
Narrower Range, No Association with Poverty
Title I per Poverty Pupil and Census Poverty Corrected for CWI
14
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
DV = T 1 Aid per Pupil Coef. Std. Err. P>t Coef. Std. Err. P>t Coef. Std. Err. P>tCensus Poverty Rate 1790.278 7.840 * 1806.494 7.743 * 1829.312 7.955 *NCES ECWI 232.741 5.286 * 209.890 5.244 * 214.252 5.239 *Structure and Size
Unified K-12 29.315 2.296 * 23.932 2.265 * 18.929 2.296 *Enroll Under 100 200.548 18.777 * 185.964 18.478 * 177.950 18.437 *Enroll 100 to 299 117.025 7.279 * 107.345 7.168 * 100.742 7.172 *Enroll 300 to 599 73.280 5.063 * 64.659 4.990 * 59.808 4.995 *Enroll 600 to 1199 48.905 3.327 * 39.108 3.290 * 35.812 3.296 *Enroll 1200 to 1499 34.895 4.156 * 25.944 4.101 * 23.855 4.094 *Enroll 1500 to 1999 82.921 3.044 * 71.523 3.026 * 70.285 3.020 *Enroll over 2000
LocaleLarge CityMidsize City -59.606 2.809 * -56.846 2.766 * -55.747 2.759 *Small City -63.041 2.821 * -67.587 2.786 * -66.269 2.779 *Large Suburb -87.539 2.161 * -87.384 2.130 * -87.736 2.127 *Midsize Suburb -77.617 3.818 * -79.156 3.759 * -76.714 3.752 *Small Suburb -87.032 4.500 * -85.724 4.428 * -83.388 4.418 *Fringe Town -77.631 3.794 * -79.711 3.735 * -77.885 3.726 *Distant Town -66.419 3.539 * -71.271 3.488 * -69.366 3.480 *Remote Town -49.204 3.860 * -53.516 3.801 * -51.946 3.791 *Fringe Rural -90.207 3.041 * -93.776 2.998 * -91.938 2.991 *Distant Rural -89.690 3.469 * -92.968 3.417 * -88.622 3.422 *Remote Rural -40.078 4.726 * -40.892 4.650 * -37.806 4.641 *
Year = 2005 7.518 1.519 * 4.983 1.498 * 5.209 1.497 *Year = 2006 -1.664 1.553 -13.319 1.583 * -11.718 1.588 *Effort 3348.274 118.756 * 2766.342 127.063 *NAEP 07 1.231 0.107 *Constant -285.805 7.980 * -360.985 8.384 * -680.485 28.787 *R-squared 0.708 0.714 0.716
Model 1 - Baseline Model 2 - Effort Model 3 - Effort&NAEP
*p<.05
Really hideous regression model of district level T1 Aid
15
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
$0
$200
$400
$600
$800
$1,000
$1,200
Unified K-12 vs.
Non-K-12
EnrollUnder 100
Enroll 100to 299
Enroll 300to 599
Enroll 600to 1199
Enroll1200 to
1499
Enroll1500 to
1999
Enrollover 2000
Dif
fere
nce
in T
itle
1 A
id
T1 per Pov
T1 per Pupil
Title I resources per poverty pupil are actually much higher in small districts
Regression Based Estimates of T1 per Pov. Pupil in Smaller Districts (rel to larger)
16
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
-800
-700
-600
-500
-400
-300
-200
-100
0
Dif
fere
nce
in T
itle
1 A
id
T1 per Pov
T1 per Pupil
Regression Based Estimates of T1 per Pov. Pupil by Locale
17
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
-500
-300
-100
100
300
500
700
900
1,100
1,300
1,500
Wyo
min
g N
orth
Dak
ota
Del
awar
e A
lask
a M
assa
chu
sett
s M
aryl
and
N
ew Y
ork
M
ain
e N
ew H
amp
shir
e R
hod
e Is
lan
d
Con
nec
ticu
t N
ew J
erse
y S
outh
Dak
ota
Wes
t V
irgi
nia
C
alif
orn
ia
Mon
tan
a P
enn
sylv
ania
O
rego
n
Vir
gin
ia
Flo
rid
a I
llin
ois
Was
hin
gton
L
ouis
ian
a K
ansa
s O
kla
hom
a A
lab
ama
Mic
hig
an
Wis
con
sin
S
outh
Car
olin
a I
dah
o N
evad
a M
inn
esot
a M
issi
ssip
pi
Mis
sou
ri
New
Mex
ico
In
dia
na
Col
orad
o I
owa
Neb
rask
a T
enn
esse
e U
tah
A
rkan
sas
Ari
zon
a T
exas
Ove
r/U
nd
er F
un
din
g of
Tit
le 1
T1 per Pov
T1 per Pupil
States with High Marginal Title I Funding Differences (excesses)
States with Low Marginal Title I Funding Differences (Deficits)
Regression Based Estimates of Winners/Losers (Model Residuals)
18
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
MontanaSouth Dakota
North Dakota
Idaho
Wyoming
ArkansasIowa
Mississippi
Maine
Kansas Oklahoma
NebraskaNew Mexico
West Virginia
LouisianaAlabama
MissouriIndiana
Oregon
Arizona
South Carolina
Tennessee
Florida
Utah Texas
Colorado
New Hampshire
Minnesota
Alaska
Pennsylvania
Delaware
Wisconsin
Virginia
MichiganWashington
Rhode Island
Nevada
Illinois
California
MassachusettsMaryland
Connecticut
New York
New Jersey
-500
050
010
0015
00S
tate
Fix
ed E
ffect
.1 .15 .2 .25 .3Census Poverty Rate
State Fixed Effect per Pov Pupil Fitted values
To some extent, the rich are getting richer (r-squared = .17)
19
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Which states are more deserving?
And How bad are the current EFIG Equity Indicators?
20
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
AK
AK
AK
AK
NJ
NJ
NJ
NJ
LA
LA
LA
LA
MN
MN
MN
MN
OHOH
OH
OH
UTUT
UTUT
ININ
ININ
SCSC
SCSC
CTCT
CTCT
MAMA
MAMA
TN TN TN TNNM NM
NMNM
OROR
OROR
KY KY KY KYWV WV WV WV
FL FL FL FL
DE DE DE DE
AR AR AR ARGA GA GA GA
OK OK OK OK
RI RI RI RI
CA CA CA CANE NE NE NEIA IA IA IA
MS MS MS MS
TX TX TX TX
MI MI MI MI
AZ AZ AZ AZID ID ID ID
MO MO MO MOWA WA WA WA
COCO CO CO
NC NC NC NC
NVNV
NVNV
NYNY
NYNY
ALAL
ALAL
MDMD
MDMD
WIWI
WIWI
VT
VTVT
VT
KSKS
KSKSMT
MTMT
MT
SDSD
SDSD
WY
WY
WYWY
VAVA
VAVA
PA
PA
PA
PAIL
IL
IL
IL
ME
ME
ME
ME
ND
ND
ND
ND
NH
NH
NH
NH
$4,500
$6,500
$8,500
$10,500
$12,500
$14,500
$16,500
$18,500
0% Poverty 10% Poverty 20% Poverty 30% Poverty
Poverty (Census) Rate
Pre
dic
ted
Sta
te &
Loc
al R
ev. P
er P
up
il
High Spending, Progressive State
High Spending, Regressive States
Low Spending States
Regression Based Estimates of State & Local Revenue Distributions
21
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
Florida Georgia
Idaho
Illinois
Indiana
IowaKansas
Kentucky
Louisiana
MaineMaryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North CarolinaNorth Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode Island
South Carolina
South DakotaTennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
6000
8000
1000
012
000
1400
016
000
Sta
te &
Loc
al R
ev. p
er P
upil
.02 .03 .04 .05 .06Effort Index
Pred. SlocRevPP at 20% Pov Fitted values
State Effort & High Poverty Spending Are Associated (r = .527)
Low effort, low spending states
Relating Effort and Relative Adequacy
22
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
$9,000
$10,000
$11,000
$12,000
$13,000
$14,000
$15,000
$16,000
$17,000
$18,000
$19,000
0% Poverty 10% Poverty 20% Poverty 30% Poverty
Poverty (Census)
Rev
. p
er P
up
il (
pre
dic
ted
)
NJ State and Local
NJ with Title I
PA State and Local
PA with Title I
Supplement effect of Title 1 aid in progressive state (NJ)
Supplant effect in severely regressive state (PA)
Supplementing in NJ – Supplanting in PA
23
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
$5,500
$6,000
$6,500
$7,000
$7,500
$8,000
0% Poverty 10% Poverty 20% Poverty 30% Poverty
Poverty (Census)
Rev
. per
Pu
pil
(pre
dic
ted
)
TN State and Local
TN with Title I
AZ State and Local
AZ with Title I
UT State and Local
UT with Title I
Supplant effect in regressive state (AZ)
Supplement effects in low spending, progressive (inconsistent) states
Supplanting in Arizona
24
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
EFIG
• 2) EFFORT FACTOR-• (A) IN GENERAL- Except as provided in subparagraph
(B), the effort factor for a State shall be determined in accordance with the succeeding sentence, except that such factor shall not be less than 0.95 nor greater than 1.05. The effort factor determined under this sentence shall be a fraction the numerator of which is the product of the 3-year average per-pupil expenditure in the State multiplied by the 3-year average per capita income in the United States and the denominator of which is the product of the 3-year average per capita income in such State multiplied by the 3-year average per-pupil expenditure in the United States.
25
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
EFIG• (I) IN GENERAL- For each State, the Secretary shall compute a
weighted coefficient of variation for the per-pupil expenditures of local educational agencies in accordance with subclauses (II), (III), and (IV).
• (II) VARIATION- In computing coefficients of variation, the Secretary shall weigh the variation between per-pupil expenditures in each local educational agency and the average per-pupil expenditures in the State according to the number of pupils served by the local educational agency.
• (III) NUMBER OF PUPILS- In determining the number of pupils under this paragraph served by each local educational agency and in each State, the Secretary shall multiply the number of children counted under section 1124(c) by a factor of 1.4.
• (IV) ENROLLMENT REQUIREMENT- In computing coefficients of variation, the Secretary shall include only those local educational agencies with an enrollment of more than 200 students.
26
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Alabama
Arizona
ArkansasCaliforniaColorado
Connecticut
Delaware
Florida
Georgia
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
MaineMaryland
Massachusetts
MichiganMinnesota
Mississippi
Missouri
Montana
NebraskaNevada
New HampshireNew Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode IslandSouth Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
510
1520
25F
ed E
quity
Inde
x
.6 .8 1 1.2 1.4ELC Progressiveness Indicator
Eq
uit
able
Ineq
uit
able
Regressive Progressive
Current Federal Equity Indicators Unrelated to “Progressiveness”
27
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Alabama
Arizona
Arkansas CaliforniaColorado
Connecticut
Delaware
Florida
Georgia
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
MaineMaryland
Massachusetts
MichiganMinnesota
Mississippi
Missouri
Montana
NebraskaNevada
New HampshireNew Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
Oregon
Pennsylvania
Rhode IslandSouth Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
Virginia
Washington
West Virginia
Wisconsin
Wyoming
510
1520
25F
ed E
quity
Inde
x
6000 8000 10000 12000 14000 16000ELC Adequacy Level - Predicted at 20%
Eq
uit
able
Ineq
uit
able
Low Revenue High Revenue
Current Federal Equity Indicators Unrelated to Relative Adequacy
29
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Alabama
Alaska
Arizona Arkansas
CaliforniaColorado
Connecticut
Delaware
Florida
IdahoIllinois
IndianaIowa
Kansas
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Dakota
Oklahoma
Oregon
PennsylvaniaRhode Island
South Carolina
South Dakota
Tennessee
Texas
Utah
Vermont
VirginiaWashington
West Virginia
Wisconsin
Wyoming
-100
010
020
0T
1 E
xces
s -
Bas
elin
e M
odel
.05 .1 .15 .2 .25Census Poverty Rate
Should Receive T1 Increase
Should Receive T1 Decrease
Baseline Model Estimates
30
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Alabama
Arizona
CaliforniaColorado
Connecticut
Florida
Illinois
Indiana
Louisiana
Maryland
Massachusetts
MichiganMinnesota
MissouriNew Jersey
New York
Oklahoma
Oregon
Pennsylvania
South Carolina
Tennessee
Texas
Virginia Washington
Wisconsin Alabama
Arizona
California
Colorado
Connecticut
Florida
IllinoisIndiana
Louisiana
Maryland
Massachusetts
Michigan
Minnesota
Missouri
New Jersey
New York
Oklahoma
Oregon
Pennsylvania
South Carolina
Tennessee
Texas
VirginiaWashington
Wisconsin
-100
-50
050
100
T1
Exc
ess
per
Pup
il
.1 .15 .2 .25Census Poverty Rate
T1 Excess - Effort Model T1 Excess - Baseline Model
Effort Bonus for NJ
Effort Penalty for LA
Should Receive T1 Increase
Should Receive T1 Decrease
Baseline Model Estimates with Effort Adjustment
31
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Alabama
Arizona
CaliforniaColorado
Connecticut
Florida
Illinois
IndianaLouisiana
Maryland
Massachusetts
MichiganMinnesota
Missouri
New Jersey
New York
Oklahoma
OregonPennsylvania
South Carolina
Tennessee
Texas
VirginiaWashington
Wisconsin Alabama
Arizona
California
Colorado
Connecticut
Florida
IllinoisIndiana
Louisiana
Maryland
Massachusetts
Michigan
Minnesota
Missouri
New Jersey
New York
Oklahoma
Oregon
Pennsylvania
South Carolina
Tennessee
Texas
VirginiaWashington
Wisconsin
-100
-50
050
100
T1
Exc
ess
per
Pup
il
.1 .15 .2 .25Census Poverty Rate
T1 Excess - Effort&NAEP Model T1 Excess - Baseline Model
Should Receive T1 Increase
Should Receive T1 Decrease
Baseline Model Estimates with NAEP Deficit Adjustment
32
Bru
ce D
. Ba
ker, A
ER
A 2
00
9 S
an
Die
go
Conclusions/Implications• Regression based estimates, controlling for (a) scale, (b) locale,
(c) wage variation and (d) poverty may be useful for driving Title I aid. – Additional “effort” and “performance” factors produce subtle shifts.
• Need to find better way to evaluate relative poverty and poverty effect on outcomes – Double edged sword - ideally, progressive states can offset poverty
effect on outcomes– On average, Title I really isn’t making the rich richer
• How to pressure low effort and/or regressive states to start doing the right thing, without further penalizing their kids.– Current EFIG Federal Equity Indicators relatively useless.– Quite simply, Louisiana would appear to care less. 17% of LA children
attend private school (2nd to Delaware, which also cares less).– Arizona has no interest in aiding children in poverty or LEP children
and ranks 46th in adjusted spending.