October 8, 2019

Kimpton Hotel Monaco Baltimore

Common Principles, Frameworks & Indicators

Dude, look at that scatterplot!Love this stuff!

He’s at it again Jazzman!

You’re telling me? Glad I’m done with

that Ph.D!

Evaluating School Finance Systems

1. Effort: All else being equal, more effort is better, particularly for states with less capacity. Conversely, however, states with larger economies may not require as much effort as states with smaller economies.

2. Adequacy: In light of widespread agreement that educational outcomes in the U.S. must improve, we assert, as a general principle, that allocating more resources to schools is better. However, states should also provide resources to schools that are commensurate with achieving common outcomes or improvement toward those outcomes.

3. Progressivity: States’ allocation of resources should be progressive -–i.e., districts serving more high-needs students should receive more revenue. The optimal degree of progressivity, however, might depend on factors such as the amount of inequality of education outcomes (for example, states with large achievement gaps might allocate resources more progressively)

AL

AK

AZ

AR

CA

CO

CT

DE

FL

GA

HI

ID

IL

INIA

KS

KYLA

MEMD

MA

MI

MN

MS

MO MT

NE

NV

NH

NJ

NM

NY

NC

NDOH

OK

OR

PARI

SC

SDTN

TXUT

VT

VA

WAWV

WI

WY

$5,000

$7,000

$9,000

$11,000

$13,000

$15,000

$17,000

$19,000

$21,000

$23,000

2.50% 3.00% 3.50% 4.00% 4.50% 5.00% 5.50% 6.00% 6.50%

Pred

icte

d St

ate

& L

ocal

Rev

enue

at 2

0% P

over

ty

Education Spending as a Share of Aggregate Personal Income

States with Higher Effort Spend More Per Pupil

AL

AK

AZ

AR

CA

CO

CTDE

FL

GA

HI

IDIL

IN

IA

KS

KYLA

ME

MDMA MI

MN

MS

MO

MT

NENV

NH

NJ

NM

NY

NC

ND

OH

OK

OR

PARI

SC

SD

TN

TX

UT

VT

VA

WA

WV

WI

WY

60%

65%

70%

75%

80%

85%

90%

2.50% 3.00% 3.50% 4.00% 4.50% 5.00% 5.50% 6.00% 6.50%

Teac

her W

ages

as a

Per

cent

of N

on-T

each

er W

ages

(Age

45)

Education Spending as a Share of Aggregate Personal Income

Teacher Pay is More Competitive in States with Higher Effort

AL

AK

AZ

AR

CA

CO

CT

DEFL GA

HI

ID

IL

IN

IA

KS

KY

LAME

MD

MA

MI

MN

MS

MOMT

NE

NV

NH

NJ

NM

NY

NC

ND

OH

OK

OR

PA

RI

SC

SD

TN TX

UT

VT

VA

WA

WV

WI

WY

4

5

6

7

8

9

10

0.025 0.03 0.035 0.04 0.045 0.05 0.055 0.06 0.065

Teac

hers

per

100

Pup

ils

Education Spending as a Share of Aggregate Personal Income

States with Higher Effort have More Robust Staffing Ratios

Progressivity

1. Substantial progressivity: The ratio of adjusted state and local revenue in higher-poverty districts (10, 20, or 30 percent poverty) to that of the lowest-poverty districts (0 percent poverty) within a given state.

2. Systematic progressivity: The correlation between revenue and poverty (labor market centered) among all districts within a given state.

Generating Comparable Revenue Measures

Describing State School Finance Systems

Describing State School Finance Systems

What about State v. Local?

• The system of state school finance is inclusive of state and local funding, because state law and policy shapes both.

• State aid funding can be highly progressive while the systemof education funding in a state is flat or regressive.

• Adequacy does not hinge on the source of funding.

• Effort, however, can be important to review by state versus local due to traditional generators of revenue.

Consistent Frameworks for “Within” vs. “Between”And for sorting out how charter school expansion influences within-district equity/progressivity

Model 1 Model 2 Model 3 Model 4 coef se coef se coef se coef se

Percent Low Income 605.577 739.951 3,177.208*** 675.869 525.769 626.455 -79.205 541.614

Year (2013 = Base)Year = 2014 178.255 252.892 190.666 219.900 228.875 190.868 480.605*** 163.251Year = 2015 178.309 255.645 245.487 222.364 288.304 193.030 506.272*** 165.516

Grade Range Distribution% school enrollment in grades 6 to 8 2,379.914*** 344.366 953.773*** 318.295 409.458 275.518% school enrollment in grades 9 to 12 3,472.449*** 284.892 2,017.058*** 271.230 1,798.147*** 231.930

Other Student CharacteristicsPercent ESL 231.429 1,051.169 554.414 892.013

Percent Special Education 19,996.225*** 1,553.121 18,665.847*** 1,415.741

% Students with Disabilities that are Non-Severe Disabilities -974.388 603.428

Intercept 12,487.185*** 665.230 9,111.236*** 634.154 8,707.116*** 551.236 10,070.596*** 765.302

Number of observations 520 520 520 514

R2 0.003 0.249 0.437 0.486

note: *** p<0.01, ** p<0.05, * p<0.1

Modeling Baltimore City Schools “Progressivity” of per Pupil Spending 2013-2015

Incomplete Models/Analyses Produce Erroneous Results!

WITHIN NEW YORK CITY

Spending across schools is progressive with respect to low income shares!

Spending is predictable as a function of rational factors (and in the “right” direction)

ACROSS NEW YORK STATE DISTRICTS

Spending across districts is highly regressive with respect to child poverty rates!

Spending is not very predictable as a function of rational factors (or in the “right” direction)

We must use appropriate models to distill progressiveness both across and within districts

• When we model per pupil spending across schools, not treating charters as a special subset, it appears that school spending is marginally lower in schools with higher low income shares (regressive)

• When we treat charter schools as a separate subset, we find that they, on average spend marginally more than otherwise similar district schools (while serving generally lower need populations). This creates the overall “regressive” pattern. District schools without charters are marginally “progressive”

Modeling Per-pupil Spending in Baltimore City

$2,000

$3,000

$4,000

$5,000

$6,000

$7,000

$8,000

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Per-

pupi

l spe

ndin

g

Low-income share

Free/Reduced Wt CBSA (Core Based Statistical Area) Fort Worth Dallas

Rearranging Deck Chairs in Dallas

To Summarize1. Cross-district “progressiveness” should be characterized by a regression model which includes:

Census Poverty Rates (as measure for determining progressiveness)• District/Regional Cost Factors

o Economies of Scale (enrollment size groups)o Population Sparsity (if available)

o Grade Ranges Servedo Regional Wage Variation

• Student Factorso Children with disabilities shares (ideally by two groups relative to severity)

o English Language Learners

2. Cross-school “progressiveness” should be characterized by a regression model which includes:School enrolled low-income shares (as measure for determining progressiveness)

• School Cost Factorso Grade Ranges Served (by % in grades pk-5, 6-8 & 9-12)

• Student Factorso Children with disabilities shares (ideally by two groups relative to severity)

o English Language Learners

Cutting through the clutter

Relationships Among Indicator Types

Other Indicators and Indicator Systems

1. Ed Trust• Poverty & Racial Funding Gaps (variation on progressiveness)• Corrected for competitive wage variation

2. Urban Institute• Progressiveness ratio (poverty)

• Corrected for competitive wage variation

3. Education Week

(wage adjusted & arbitrary pupil need weights)• Equity indicators

o Coefficient of Variation (raw variation in per pupil spending, statewide, as % of mean)o Restricted range ratio (ratio of 95%ile spending to 5%ile)o Fiscal Neutrality Index (relationship between wealth and per pupil spending)

• Adequacy / Spending Indicatorso McLoone Index (ratio of average of lower half spending to median)

o Spending Index • % Students in Districts above National Average PPE

Column1

Poverty Gap ($) Ed Trust

Poverty Gap (%) Ed Trust

Racial Gap ($) Ed Trust

Racial Gap (%) Ed Trust

Progressiveness ($) Urban Institute

Neutrality - Ed Week

Coefficient of Variation - Ed Week

Restrict Range - Ed Week

Poverty Gap (%) Ed Trust 0.9512

Racial Gap ($) Ed Trust 0.6847 0.6013

Racial Gap (%) Ed Trust 0.6187 0.5742 0.9719Progressiveness ($) Urban Institute 0.8009 0.6982 0.7193 0.6681

Neutrality - Ed Week -0.0776 -0.0395 0.036 0.0134 -0.1593Coefficient of Variation - Ed Week 0.0314 0.0363 -0.0364 -0.0565 0.1364 0.0414Restrict Range - Ed Week -0.056 -0.1311 0.0749 0.0824 0.2011 -0.0644 0.4628Progressiveness Ratio - SFID 0.508 0.5824 0.0832 0.0521 0.3002 0.0254 0.1885 -0.1731

Correlations Among Equity Indicators

Column1Effort (Ed Week)

GSP Effort (SFID)

Aggregate Income Effort (SFID)

McLoone Index (Ed Week)

Spending Index (Ed Week)

Current Spending as % of Target (SFID)

Current Spending Gap to Target (SFID)

Predicted Spending at 20% Poverty (SFID)

GSP Effort (SFID) 0.4884Aggregate Income Effort (SFID) 0.2803 0.8597McLoone Index (Ed Week) -0.4008 -0.3257 -0.1805Spending Index (Ed Week) 0.303 0.4555 0.5548 -0.2271Current Spending as % of Target (SFID) 0.339 0.4433 0.4659 -0.3096 0.7145Current Spending Gap to Target (SFID) 0.3146 0.4319 0.4767 -0.2915 0.7263 0.9813Predicted Spending at 20% Poverty (SFID) 0.3287 0.5727 0.6385 -0.3436 0.8185 0.8921 0.8949

Current Outcomes as Ratio to Mean (SFID) 0.1737 0.1764 0.1313 -0.2453 0.4746 0.7513 0.7489 0.5697

Effort and Adequacy Measures

Reducing interstate inequality

Toward an Adequate System for All

A Word on Inefficiency

Inefficiency

CostPrice

• Inefficiency effectively increases the price of a good.• Little evidence that inefficiencies exist on scales

similar to adequacy or equity gaps.• It has to be yes and, not a reason to fail to pursue

adequacy and equity.• Good policy understands what increases costs and

supports those districts/schools and minimizes inefficiencies.

Adequacy

Inefficiency

Spending

CostMeasured Student

Outcomes

Student PopulationInput Prices

Structural/Geographic Constraints

Efficiency Controls:

Fiscal capacity, competition, &

public monitoring

Inefficiency

Spending

CostMeasured Student

Outcomes

Student PopulationInput Prices

Structural/Geographic Constraints

Efficiency Controls:

Fiscal capacity, competition, &

public monitoring

Economies of Scale (Size)Population Sparsity

Regional Wage Variations

Social ContextPoverty (Concentration & Density)

Individual NeedLanguage ProficiencyDisability

coef seOutcome Index 8.093*** 0.654Education Comparable Wage Index 0.617*** 0.034Adjusted Poverty Rate 2.572*** 0.190State Mean Centered SWD Rate 2.153*** 0.125% ELL 0.876*** 0.103% Enrollment in Pre-k -0.004 0.128% Enrollment in K -0.465** 0.190% Enrollment in Middle Grades -1.636*** 0.145% Enrollment in Secondary Grades -0.328*** 0.075Less than 100 Students 0.378 0.279101 to 300 Students 0.324*** 0.048301 to 600 Students 0.254*** 0.033601 to 1200 Students 0.153*** 0.0301201 to 1500 Students 0.112*** 0.0391501 to 2000 Students 0.126*** 0.043Log of Population per Square Mile -0.024*** 0.006enroll_under100_x_density 0.009 0.068enroll_101to300_x_density -0.000 0.011enroll_301to600_x_density -0.007 0.007enroll_601to1200_x_density -0.003 0.006enroll_1201to1500_x_density -0.003 0.008enroll_1501to2000_x_density -0.007 0.008Unified K12 District 0.105*** 0.020% Revenue from State and Federal Sources 0.032 0.034Herfindhal Index - Enrollment -0.083*** 0.030% Population between 5 & 17 yrs of age 0.012 0.086year==2010 -0.047*** 0.005year==2011 -0.087*** 0.008year==2012 -0.100*** 0.008year==2013 -0.114*** 0.009year==2014 -0.045*** 0.006year==2015 -0.042*** 0.008_cons -36.127*** 3.664Number of observations

IV Reg Grade 3-8

80,503

Student Needs

Grade Range Distribution

Scale and Sparsity

! Higher outcomes cost more to achieve !

Regional labor costs

Current spending (2013-2015) as % of

“cost” of achieving national average outcomes (red =

lower, green = higher)

Current outcomes (2013-2015) with

respect to national average outcomes

(red = lower, blue = higher)

The Road Ahead

1. We need to rally our forces around core principles. Many (most) of us largely agree on the core principles: • Proper funding is a necessary condition for educational success:

Competitive educational outcomes require adequate resources, and improving educational outcomes requires additional resources.

• The cost of providing a given level of educational quality varies by context: Equal educational opportunity requires progressive distribution of resources, targeted at students and schools that need them most.

• The adequacy and fairness of education funding are largely a result of legislative policy choices: Good school finance policy can improve student outcomes, whereas bad policy can hinder those outcomes.

2. We need to clear the clutter when it comes to indicators of state school funding systems

• Indicators should address core principles:• Indicators involving spending measures must sort out “good” (equitable,

cost-based) variation from “bad” (inequitable) variation • Address regional variation in labor costs• Address economies of scale and population sparsity• Address student need related cost factors

3. We need to work as a team to help state policymakers understand the relationships among our indicators, and what they say collectively about state school finance systems!