Financial Transparency Working Need. Data. Now. How good is good enough? March 28, 2017 By: Marguerite Roza Edunomics Lab at Georgetown University

Communicating During the Call To request to be unmuted, please use the hand raise tool Use the comment box to ask questions or message the group bscpcenter.org

Agenda Need. Data. Now. How good is good enough? Examples Colorado Developing a framework for data standards Wisconsin bscpcenter.org

bscpcenter.org

Need. Data. Now. 2. SEA has SLFS (or similar) data by school 3. SEA has real salaries/benefits of personnel with location 4. SEA does not yet have access to financial information by school: 1. SEA has a chart of accounts (COA) with a field for location CO DE, FL, HI, MA, ME, MS, OH, RI, DC, MD, WY, OR, NE, MD CT. IL ND, VA, AZ, SD, MO, TN Already have data (lucky you). Your SEA already has data (be it expenditure data, salary data, or SLFS data). Next steps: run early analyses of the data and consider allocation rules for centrally assigned costs. Ask for electronic files. Maybe LEAs in your state are using location codes (even if not consistently) and you could ask for the raw data files and do any work of integrating & calculating the PPE from those files. Files to collect might include expenditure data and/or personnel files. Issue a survey. Perhaps electronic files won’t yield anything of value or are so inconsistent as to warrant SEA analysis. A third option is to issue a SURVEY to collect information from your districts. One tested survey instrument is the SLFS survey, but you could issue your own.

How will your state obtain school level data How will your state obtain school level data? Use chat box to indicate which option # applies to your state: Already have data (lucky you). Your SEA already has data (be it expenditure data, salary data, or SLFS data). Next steps: run early analyses of the data and consider allocation rules for centrally assigned costs. Ask for electronic files. Maybe LEAs in your state are using location codes (even if not consistently) and you could ask for the raw data files and do any work of integrating & calculating the PPE from those files. Files to collect might include expenditure data and/or personnel files. Issue a survey. Perhaps electronic files won’t yield anything of value or are so inconsistent as to warrant SEA analysis. A third option is to issue a SURVEY to collect information from your districts. One tested survey instrument is the SLFS survey, but you could issue your own. Not sure yet.

What portion is tracked to the school-level? How good is good enough? Already have electronic data Collect data files Issue a survey Examples Rhode Island: has common COA, rules about what is/is not coded to schools & how, collects at the state level MD AIR Study: collected electronic data files from districts (expenditure survey & personnel) to find PPE at school-level SLFS: federal survey issued to districts that collects school-level data on a subset of expenditure categories What portion is tracked to the school-level? 65-98% ~54% 37-54% What’s the burden? Minimal burden to LEAs. SEAs can do all the analysis. LEA must extract and send files. Time and resources at SEA level to clean, merge, analyze LEAs to complete survey, and SEAs to review & verify

Additional considerations Already have electronic data Collect data files Issue a survey Uniformity? Depends on whether there are some common practices re attribution or COA specifies attribution High – SEA controls framework for analysis Depends – Higher if common COA, lower if not Chance for error? Lower: LEA coding errors might exist, although patterns should be evident in the data. Medium-Low: LEA coding errors, but can be checked at SEA level Medium-High: variation in LEA interpretation & reporting Other considerations Run consistency checks across districts to explore uniformity in attribution May be a good short term strategy -- gives SEA opportunity to standardize Survey must include breakout by source of expenditures (SLFS has with and without exclusions) For all, worth triangulating w/ add’l data sources (i.e. personnel files; F-33s) bscpcenter.org

How good is good enough? Attributing costs by formula If attributing health or retirement benefits by formula, better to do so by STAFF SALARY. Pupil support and instructional support not well attributed by these formulas Source: U.S. Department of Education, Office of Planning, Evaluation and Policy Development, Policy and Program Studies Service, Exploring the Quality of School-Level Expenditure Data: Practices and Lessons Learned in Nine Sites, Washington, D.C., 2017. bscpcenter.org

How good is good enough? Attributing costs by formula Considerations on how/whether to attribute costs (vs leaving costs at LEA level) -- Amount of money involved -- Relevance for schooling -- How much error if done formulaically vs via actual costs -- Lumpiness of spending -- Frequency of costs incurred (yearly, once per decade, etc.). ERS is working on a tool for LEAs to explore attribution of central costs. Let us know (via chat box) if you’d like to be part of a webinar demonstration bscpcenter.org

Analysis of one CO school district Expenditure Files SLFS (w/o exclusions) Elementary School Middle School High School A School level Federal $675 $209 $197 N/A B S/L $6,378 $6,856 $8,463 C Sch total $7,052 $7,065 $8,659 $7,891 $6,838 $8,955 D LEA level $61 E $2,378 F LEA total $2,439 $2,586 G Grand Total $9,492 $9,504 $11,099 $10,477 $9,424 $11,541 District characteristics: 7 schools (2 ES, 1 MS, 1 “secondary”, 1 HS, 1 pre-k, 1 charter) % SL captured by SLFS for this district: 72-76% Expenditure file captured 78% of selected expenditures at the school level. SLFS data (without exclusions) captured 76% of total expenditures (F-33) SLFS data with exclusions captured 65% of total expenditures (F-33) Exclusions include: 1) expenditures from federal revenue sources other than federal funds intended to replace local tax revenues, 2) prekindergarten expenditures, and 3) special education expenditures bscpcenter.org

Possible Framework for Drafting Reporting Standards: The goal: Enable spending comparisons across schools in different states. Tier 1: Minimum Tier 2: More information Tier 3: Richest data Define student enrollment count procedures Clarify what expenditures are excluded/included or how to specify All LEA expenditures accounted for (in either school level or LEA level Specify minimum breakouts: Special ed, bilingual ed, nutrition, transportation, etc. Specify additional breakouts re objects, functions, etc. (benefits, etc.) Clarify options for attribution Align with comparable student outcomes (percentiles?) bscpcenter.org

Milwaukee example bscpcenter.org bscpcenter.org 13

Resources MD AIR Study: http://www.marylandpublicschools.org/about/Documents/OCP/Publications/MarylandStudyPublcSchoolFunding122016.pdf DoE AIR Study (Case study on 9 sites, including RI and HI) https://www2.ed.gov/rschstat/eval/title-i/quality-expenditure-data/report.pdf bscpcenter.org