Bruce D. Baker Rutgers University

1. Bruce D. Baker, AEFA 2009 Rearranging Deck Chairs in Dallas: Contextual Constraints and Within District Resource Re-allocation in Urban Texas School Districts Bruce D. Baker Rutgers University AEFA 2009 Nashville, TN

2. Bruce D. Baker, AEFA 2009 Research questions • How do aggregate resource levels and teacher characteristics of urban core districts compare with their surroundings? • How do fiscal resources and teacher characteristics vary across schools within large urban districts? • How does need and cost related variation within large urban districts fit within the broader labor market of resource variation among schools in surrounding districts? • That is, is the district a sinking ship facing potential constraints on within district re-allocation? • To what extent does existing variation in resources account for the marginal costs of providing equal educational opportunity across schools within large urban districts and between large urban districts and their surroundings?

3. Expend per Pupil Expend per Pupil Poverty Rate Poverty Rate Bruce D. Baker, AEFA 2009 Figure 1 Contextual constraints on within-district resource allocation Suburban schools - flat distribution Urban schools - regressive distribution Suburban schools - flat distribution Urban schools - progressive distribution

4. Bruce D. Baker, AEFA 2009 Data & Analyses • 3-year panel of 1409 Texas Elementary Schools in 5 major cities in 4 Core Based Statistical Areas • Why only 3 years? • To be able to estimate marginal cost model based on consistent outcome measurement. • Models • Expenditure decomposition • ln(expend) = f(size, students [se, lep, pov]) • Drop SE marginal expend • Drop small schools • Models estimated within city districts • Poverty slope projection • ln(expend) = f(Pov x District) • Excluding SE and small schools • Estimated for each CBSA, with Pov x City District interaction • How does the district’s poverty slope fit within the CBSA (and how do the intercepts align? (is the district a sinking ship?) • Cost model • Expenditure function, including outcomes (instrumented) & efficiency controls • Model estimated across all districts within 4 metros

5. Table 1 Characteristics of Urban Core and other districts sharing CBSA (2007) Bruce D. Baker, AEFA 2009

6. Figure 2 Current Operating Spending per Pupil in Urban Core and Surrounding Districts by Grade Level Bruce D. Baker, AEFA 2009

7. Figure 3 Teacher Characteristics in Urban Core and Surrounding Districts (Elementary Schools) Bruce D. Baker, AEFA 2009

8. Table 2 Models of Current Spending and Teacher Characteristics in Texas Cities Bruce D. Baker, AEFA 2009 Houston’s allocation is least sensitive to poverty among the districts in the analysis. *p<.05, **p<.10

9. Figure 4 Predicted spending by poverty for scale-efficient elementary schools Ft. Worth ISD Houston ISD Dallas CBSA Dallas ISD Houston CBSA Austin ISD San Antonio CBSA San Antonio ISD Austin CBSA Bruce D. Baker, AEFA 2009

10. Figure 5 Proximity of City Elementary Schools to Neighbors (Dallas City and CBSA) Bruce D. Baker, AEFA 2009 Dallas & Mesquite ISD Per Pupil Spending (elementary schools) Dallas & Mesquite ISD % Free/Reduced Lunch (elementary schools)

11. Bruce D. Baker, AEFA 2009 Table 3 Global Cost Model Estimated using instrumental variables, where excluded instruments include (a) the mean percent black students among all other (excl. observation) elementary schools in the same district and (b) the ratio of the elementary school aged population in the district to the adult population of the district (from Census 2000). These instruments prove to be relevant instruments, and strong ones, which also pass tests for over-identification. *p<.05, **p<.10

12. Figure 6 Predicted poverty related costs and spending in city district and CBSA Bruce D. Baker, AEFA 2009 Fort Worth Austin Dallas San Antonio Houston

13. Table 4 Factors associated with Cost Adjusted current spending Bruce D. Baker, AEFA 2009 Dallas “underfunds” schools which have more students with disabilities (relative to cost model) Houston “underfunds” higher poverty schools relative to marginal costs of equal opportunity (average outcomes) Ft. Worth and Austin achieve neutrality (statistically) with respect to marginal costs related to poverty concentrations *p<.05, **p<.10

14. Table 5 Cost Adjusted Mean School Level Expenditures Deficits in Dallas & San Antonio Parity in Houston & Ft. Worth Surplus in Austin Deficits in Dallas & San Antonio

15. Bruce D. Baker, AEFA 2009 Conclusions/Policy Implications • Austin and Ft. Worth ISD seem to be best targetting resources on the basis of school level poverty, but are also reasonably well positioned to do so. • Houston has accomplished a relatively flat distribution, whereby the Houston poverty slope is lower than the marginal cost slope. In this sense, Houston’s WSF has actually worked against the goal of providing equal educational opportunity. • Dallas appears to be resource constrained when considering only elementary school budgets. These constraints severely limit Dallas’ re-allocation options. It’s like … • Dallas appears to fund middle schools well, but also allocates less to high schools • District level finance data provide conflicting evidence regarding available resources in Dallas

16. Bruce D. Baker, AEFA 2009 Table A1 Alternative Global Cost Model with District Level Wage Adjustment Estimated using instrumental variables, where excluded instruments include (a) the mean percent black students among all other (excl. observation) elementary schools in the same district and (b) the ratio of the elementary school aged population in the district to the adult population of the district (from Census 2000). These instruments prove to be relevant instruments, and strong ones, which also pass tests for over-identification. *p<.05, **p<.10

17. Bruce D. Baker, AEFA 2009 Table B Major Urban School Districts, Relative Poverty and Current Spending Levels Data Source: Current expenditure and enrollment data from U.S. Census Fiscal Survey 2005-06. Poverty data form Small Area Income and Poverty Estimates. Labor Market Area definition from NCES Education Comparable Wage Index. Includes NCES Common Core Locale Codes for Large City (11) and for Fringe of Large City (21) only.