Recalculated: “Fab-4” School Closure Savings — it’s small
(tl;dr? see 1-page flyer of the graphs and key points)
Executive summary
Closing four schools will yield savings of $630k-$678k per year per school, substantially less than the SPS estimate of $1.5M possible savings per school. We base our numbers on official OSPI F196 and S275 data, which describe actual district expenditures and staffing compensation, in detail. For the next two years, we will likely incur from $282k-$322k per year of costs related to closing schools. This halves the savings and produces a deficit reduction of only $308k-$396k per year for the next two years.
The SPS savings estimate inflates transportation and custodian savings, resulting in a substantial overestimate of the expected savings per school closure. Contrary to the assumption that SPS is making, it is unclear if there are any transportation savings as closing a school puts more kids on buses and may decrease transportation funding in the “STARS” formula due to reduced destinations. Similarly, the district expects average savings of $365k/year per school from reduced custodial staffing requirements, when the average custodian makes $80k in total compensation. This overestimates that each of these small schools employs 4–5 full time custodians.
Previously, we showed that the 21-school closures Well-Resourced School plan had a disparate racial impact, particularly in the north end of Seattle. The current 4-school plan continues to have disparate intersectional impact to OSPI homeless, low-income, ELL, and disabled populations. We have shown that needing to “right-size” for enrollment is a continually repeated falsehood as we have 2000–4000 more students than any year in 2000–2011 (and possibly much earlier). Here, we show that the fiscal savings argument for closures is overstated.
Estimation Detail
This estimation was produced via in-depth review of Seattle Public Schools expenditures using the available public data.
We estimate the current plan to close 4 schools will reduce the budget deficit by a total of $1.2M/year to $1.6M/year — or 1.3% to 1.7% of the 94M SPS budget deficit. Looking at it another way, these closures only reduce the annual operating budget by 0.1%. At this amount, the savings from closing a school is nearly the same as the average total compensation for just 4 of the 89 “Directors or Supervisors” allocated to the Central Administration in the 2022–2023 OSPI S275 personnel book.
Ignoring the one-time closing costs, and looking at it from a per-building perspective, closures would net an average ongoing expenditure reduction in the range of $630k/year to $678k/year of savings per building (total $2.5M to $2.7M across 4 buildings).
This is in contrast to the $1.4M/year per building (total $5.7M/year across 4 buildings) of savings the district is claiming is possible in their Oct 24th Estimate (see later).
We estimate these savings ($490k per school — from 72% to 78% of all savings) will come from eliminating social workers, counselors, librarians, operational staff, and replacing a principal with an assistant principal.
The non-staff portion of the savings come from building fixes, ongoing maintenance, and minimal utility expenditures. It is important to note that a higher estimate is caused by deferred maintenance/replacement of the school building (these buildings have been repeatedly left off of BEX levy lists for years). One has to wonder how much of this non-staff portion of savings would go away if the next BEX levy was adjusted to replace these school buildings.
Additionally, closing schools incur large one-time costs in the short run and have a history of causing significant loss of enrollment in closed schools. Adjusting the numbers from 2003 dollars, for the next 2 years, we will incur additional expenditures of $282k-$322k/year per school. This makes the short-term deficit reduction only $308k-$396k/year per school (a total of $1.2M to $1.6M across 4 buildings) This is important if the short-term budget deficit reduction is the primary motivation for closing buildings now.
Additionally this estimate, as well as those produced by the district, are done expecting no enrollment loss so the real savings are likely lower than the estimates. Future closures targeting buildings in better condition should also expect even lower savings.
This analysis shows that there is no strong fiscal reason for closures.
Can you show your math?
Sure. It’d be irresponsible not to.
(Links: spreadsheet of summary, and a 1-page PDF of graphs)
We used the 2022–2023 OSPI “F196” data, as this is the most recent available public data on actual expenses. Each district in the state has to report their actual expenditures (as opposed to what is projected in the budget) every school year. These are termed the “actuals.” Using the F196 data to examine SPS makes the raw figures in this analysis very precise.
Expenditures in the district are either assigned to an individual school or to the “District Office”. Expenditures coded to the “District Office” summed to $407M — 36% of the $1,124M Total Expenditures. It includes everything from “Settlements and Judgements Against the School Dist(rict)” to all food for school cafeterias. Some of these expenditures are divided across schools as services and some are purely allocated to Central Administration activities. We nicknamed this bucket the “Moshpit”.
To calculate the per school savings from each closure, we added together savings from both the individual school budgets and savings from the Moshpit. The result is a picture line the next graph.
We will start our explanation with expenditures assigned to each individual school as that is a straightforward analysis.
Expenditures assigned to each individual school.
Here is a summary of total expenditures attributed to each of the 4 targeted schools (categories match p. 44 of the 2022–23 Budget Book).
The vast majority (≥ 99%) of expenditures are staffing related (e.g. salary and benefits, contracting education professionals, additional programs needed for students, etc.). There is a very small amount left that is non-staff spending.
The analysis below starts with the staff-related spending as that is the largest bucket, and then looks at the non-staff fixed costs.
Overall savings at each individual school
Overall, we estimate a total of $297k/year saved in each closed school’s budget from reductions in utilities and elimination of staffing roles (principals, librarians, music/art/PE teachers, and administration).
The savings calculation includes the additional General Education teachers that will be allocated to the new school, but excludes the cost savings of maintenance, building Operations staff, and nurses as those three items are accounted for in the Moshpit.
The next two sections outline the savings in utilities and staff.
Staff savings from the individual school expenditures
Average cost savings from staff reductions amounts to only $271k per school.
SPS’s proposal removes Principals, Librarians, Music/Art/PE Teachers, and Administrative Secretaries. They replace them with a smaller number of Assistant Principals and General Education instructors (Additional Analysis for Preliminary Recommendations of 2025–26 School Closures). Changes to Nursing staff are excluded in this section as they are accounted for in the Moshpit calculation later.
The following table shows the expenditure reductions of all staffing changes summed together across four schools multiplied by the average total compensation (salary, benefits, insurance) of the positions being removed. The average is taken over the actual positions in the 4 closed schools:
The following table shows the new expenditures in the merged schools and is calculated using the average total compensation for the position across the entire district. This yields a slightly higher cost for Basid Ed by a few thousand a year.
All compensation data is taken from the State’s S275 personnel report data files which gives a detailed breakdown of staff allocation to buildings, duty title, and compensation. Note that while the S275 data is publicly available and shows exact roles per school, we are choosing to present in aggregate as that is sufficient for the argument. Each of these line items represents a human in our community and though it is necessary to examine them, attempts should be made to protect as much privacy as possible. A link to an identity obfuscated SQL database of the S275 is provided at the end.
Note also that the S275 is not filed at the same time as the F196 and updates are not always captured equally in both versions. Also, the S275 does not capture well the changes caused by workers compensation, overtime, unexpected leaves, departures, or other such issues. This leads to some discrepancies in numbers between the total of the S275 line items and what you see in the F196 line items but the overall result still holds.
From the table, we can conclude about $271k savings per school.
Non-staff savings at each individual school
Closure of the 4 targeted schools only yields $26k/year of savings per building in terms of non-staffing based operating costs, exclusive of maintenance, operations costs, and the larger portion of utilities that are charged to the Moshpit.
To calculate, we start with the total non-staff expenditures in the individual school and exclude items that will continue irrespective of school closure. We find that the vast majority of the costs follow the student.
Here is the starting table showing all non-staff expenditures which we will then whittle down until just $26k/year:
In the first three categories (Teaching, Principal Office, Teaching Support) the non-staffing expenditures consist of:
- General Supplies
- Student Transportation Svcs
- Travel, Entrance Fees
- Some scattered small items
These can be dropped from the estimate of savings from a closure because the costs likely follow the student to the new school or are non-recurring expenditures.
The Uncategorized expenditures in Sacajawea and Stevens includes interest on short-term debt which also does not go away with closure and can be dropped from savings estimates.
The Central Administration expenditures are too small to matter so they are dropped.
The remaining category of Other Support Activities averages to $43k/year per build.
On average, Electricity and Utilities Service make up 83% of this cost, the rest being “General Supplies”. (Note that the Moshpit, confusingly, also has a large utility line item that we address later.)
Per school utilities can be reduced in a closed school building but not eliminated as a commercial HVAC system must keep running to prevent system failure. The building also needs power for minimal security systems and lights. It’s hard to know the exact savings from reducing utility usage at a closed school building. The district in its Oct 24th estimate assumed 40% of the cost remained even in a vacant building so we will also use this figure for the entire Other Support Activities bucket which yields a savings of $26k/year per building.
By that logic we will assume a $26k/year reduction of non-staffing expenditures.
Adding together the $271k/year of staff and the $26k/year of non-staff expenditure reductions gives us the average $297k/year per building for each individual school budget.
Expenditures in the District Office “Moshpit”
Closure of 4 schools yields an estimated range of $333k/year to $381k/year reduction per building. Expenditure reductions come from $140k/year in custodial services; $47k/year in food services; $21.8k/year in grounds maintenance; $11.8k/year in nursing staff; and $113k/year to $162k/year in building fixes, utilities, and general supplies reductions
In Seattle Public Schools, one-third ($407M out of $1,121M) of the district-wide budget has been assigned to the “District Office,” which in frequently cited reports such as OSPI Report Card, get evenly distributed out to all students leading to misleading figures. For example, in 2022–2023, the “District Office” bucket accounted for more than $8k of the $22k average expenditure per student. In other words, we have to dig deeper to understand the sources of a full one third of the average per-student expenditure.
This $407M District Office Moshpit includes everything from “Settlements and Judgements Against the School Dist” for lawsuits, all food for school cafeterias, Speech Pathologists, Bilingual Transition Teachers, School Nurses, and Custodial staff, for example.
This means, frequently, over one-third of the costs attributed to each school by the OSPI Report Card do not reflect actual usage by that school; a student that is fluent in English is not receiving Bilingual Transition Teachers yet the expenditure is still partially attributed to them in the OSPI Report Card.
Understanding the reductions means cutting apart the line items as best as possible and then further cross referencing staffing data to break apart the compensation expenditures.
As before, we split the analysis into staff-based reductions and non-staff based reductions. The non-staff savings will require some guess-work as we do not have per-school allocation of repair costs.
Note: while this large Moshpit of funds makes it difficult to understand where district expenditures are actually going (likely even for their own staff), it seems more to be a result of messy accounting rather than anything improper.
Moshpit Staff Expenditure Reductions
We estimate $220k saved per school per year due to staffing reductions allocated in this budget.
Reductions come from eliminating $140k/year of custodial staff (1.75 FTE) per school; $47k/year of Food service staff (0.54 FTE); $21.5k/year Grounds Maintenance (0.25 FTE); and $11.8k/year in Nursing (0.1 FTE).
From the S275, the average total compensation for Nurses, custodial staff, and ground maintenance staff (listed in S275 as “Service Worker” assigned to the “Operations of Buildings” or “Grounds Maintenance” activities) are:
- Custodian — $80k/year
- Food Service worker — $86k/year
- Ground Maintenance stuff — $87k/year
- Nurse: $118k/year
For context, in 2022–2023 the S275 lists a headcount of 352 district-wide custodial staff, 154 food service staff, 32 ground maintenance staff, and 78 Nurses. These are likely overcounts as the S275 will double-count a departure + rehire but the numbers should give a sense of scale. These numbers also are headcounts. Many workers — especially food service staff, do not work anywhere near 1 FTE.
For our estimate, we note that Sanislo had one full-time day and one full-time night custodian and assume other small schools have 2 custodians each. We also know from the 2007 closure experience we know that the district had to retain one custodian to fend off property damage such as copper theft at the newly vacant buildings. Thus, we can assume 8 custodial assignments removed, and one created. This yields a net reduction of 7 custodians over 4 school closures eliminating 1.75 custodians per school.
For food service, parents at each of the targeted schools cite only having one food service employee each day. From the S275 data, the Service Workers assigned to “Operations — Food Service” average 0.544 FTE per duty assignment (in fact there is only one in the entire district that has a full 1 FTE; everyone else is 0.73 or lower).
For grounds maintenance, since ~30 grounds maintenance staff cover all 108 school + supporting buildings, we can assume each school closure eliminates 0.25 staff grounds maintenance staff per school.
For nurses, per the Additional Analysis for Preliminary Recommendations of 2025–26 School Closures, only 0.4 Nurse positions will be eliminated from the Moshpit when closing 4 schools freeing 0.1 Nurse position savings per school.
Thus the net Moshpit staffing reduction is (1.75 * 80k + .544 * 86k + .25 * 87k + .1 * 118k)/year = 220k/year.
Note that Building and Property Security has 15 employees that probably represent the security guards assigned to each of the district’s 8 “comprehensive” high schools. These will not be affected by elementary closures.
There are also 14 Utilities employees. It’s unclear what this staff is for but the number of people is too small to scale by the number of schools closed. They are thus excluded from the savings calculation.
Transportation employees will not be reduced and in fact will likely increase as we are displacing a few hundred students from their walk zones and moving them to further schools. However, since the STARS formula reimburses General Education buses, it is quite possible this is deficit neutral.
Note that the transportation savings assumption here is critically different from the estimates that the district is using. See the later section on divergence from district estimate.
Moshpit Non-Staff Expenditure Savings
We estimate from $113k/year to $162k/year per school in savings from maintenance and additional utilities. Variations come from different models for how much more expensive these 4 buildings are than average to maintain yearly.
The reductions come from $49k/year to $97k/year of building fixes; $42k/year of Utilities; $12k/year in cleaning services; and $11k/year of General Supplies used in maintenance + operations.
For the numbers below, we start with SPS’s Oct 24th estimate that the average maintenance expenditure per school is $49k/year. We valid that against the average building fix expenditure across all schools. Then we make a guess on a reasonable upper-bound for the closed schools based on our knowledge that the schools have a lot of deferred maintenance.
From the F196 data, building fixes sum to $6.2M/year and consist of 1.3M/year Capital Outlay and 5.1M/year Purchased Services (mostly contractors for repair and renovation).
This is totalled across all 108 schools and an unknown number of support buildings or things like Memorial Stadium. Conservatively, we will ignore the unknown buildings and divide the costs by 108, which will even exclude the John Stanford Center incorporating its share of non-staff costs. This yields an average cost of $57k/year which is higher than the SPS Oct 24th estimate but in the right ballpark.
To create a conservative upper bound for maintenance, we double the SPS Oct 24th number and estimate that building fixes cost between $49k/year and $97k/year.
Note the excess repair costs we are attributing to targeted schools are not properties of the school communities themselves. They exist because the district has left these schools off BEX levies for decades. The excess repair costs can likely be eliminated if the upcoming BEX levy was appropriately targeted to replace the existing buildings. Capital improvements via the levy do NOT affect the operating budget and deficit.
While we can consider the excess repair costs from deferred maintenance as part of the savings, it is disingenuous and short-sighted planning to believe that closure of school is the most fiscally responsible way to achieve elimination of maintenance. In reality, closure over replacement ensures we have the unnecessary continued upkeep utilities, and property insurance on a building that isn’t used at all. We will trade filling excess capacity that costs very little because it is in a modern energy efficient building for ensuring that an older vacant one will have continued costs that are just completely wasted.
Furthermore, the idea that physical plant’s expiration is driving school closure should be scoped to only saving between $49k/year and $97k/year, less than 0.1% of the Budget Deficit. These numbers make physical plant maintenance concerns an exceedingly weak fiscal argument.
Next, we examine Utilities. There is $7.4M charged to the district office for Utility Service, Natural Gas, Electricity, Communications; and $1.3M in cleaning services.
It is unclear why Utilities are split between the individual school budgets and the Moshpit. It is also unclear why there are $1.3M in cleaning services beyond what would have been handled by the $20M spent on regular 111 Maintenance Crafts and Trades Workers, 352 Operations of Building Service, and 32 Grounds Maintenance Service Workers.
But averaging both those numbers over 108 schools and using the SPS Oct 24 assumption that 40% for utilities themselves expenditure is on-going after closure we get $41k/year utility spend and $12k/year (not scaled down to 40%) of cleaning fees. This should be a large over-estimate since this includes non-school buildings as well as high schools + middle schools all of which have huge way more square footage than the targeted smaller schools. This leaves us with ~62k/year of Utilities and Cleaning services removed per closed school.
There are also $1.9M of general supplies across the Maintenance, Utilities, Operating of Buildings, and Grounds Maintenance activities. Spreading that across 108 buildings (again an overestimate due to the smaller size of the schools) yields ~11k/year per school of shared General Supplies.
Summing the above together gives a range of $113k/year to $162k/year per building.
One-time closing costs over 2 years
From the 2007 closure experience, we know that closing schools has higher than expected one-time costs. Here is the relevant snippet of the Seattle Post-Intelligencer article from November 20, 2007 titled “After 5 schools closed, 157 students left Seattle district: Only half merged into designated nearby schools”
In 2007 dollars, the cost a year and a half into the process was $927k with an expected growth to the range of $1.4M to $1.6M. Using the seattle.gov historical CPI data for “CPU-U all urban consumers, not seasonally adjusted” in the “Seattle MSA” area, we see the 2007 index was 215.656 and 2023 is 340.845 giving an adjustment factor of 1.58. Thus, we can adjust the one-time costs to $2.3M to $2.6M over 2 years, or $1.13M to $1.29M in a given year.
This number is pure overhead and will be straight up added to the budget deficit over the next two years.
Divergence from Oct 24th District Estimates on savings
SPS sent the following summary of their estimations just as this was getting ready to be published.
We incorporated some changes (eg. the 40% on-going utility) from this document into the analysis above, but there are a few notable large divergences.
First, the SPS Oct 24 estimate expects an average of $310k worth of transportation savings which seems hard to imagine as we will be expanding the number of students outside the walkzone. Furthermore, the transportation reimbursement formula gives extra money to the district for each additional destination. Playing with the Stars Funding Simulator XLS’s for the Seattle Public Schools district, removing four destinations in the simulation, increasing the number of general education students bussed by 300, and increasing the district-wide average distance bussed by 0.5 miles yielded $667k less funding per year for transit. However if the buses are fuller, it might allow SPS to be more efficient to offset more bus routes. Our guess at this time as that it is more likely that transportation changes will be deficit neutral at best. More investigation is required.
Second, the Custodial staff savings average $364k per school. The S275 data gives a total compensation for Operations Service Worker at $80k in 2022–2023 and the F196 data shows that the vast majority of custodial service spend goes to staffing. That would imply that closing each school would somehow free up 4–5 custodians full times worth of compensation meaning a closure of 4 small schools would allow 16–25 custodial headcount out of the 350 in the district to be freed. This does not make sense.
Third, the Teaching Support savings seem to be a back-of-the-envelope calculation assuming a removal of all social workers, Music/Art/PE teachers, Librarians. This is in direct contradiction to the more granular staffing headcount changes shown in their Additional Analysis for Preliminary Recommendations of 2025–26 School Closures tables. Our estimate uses the headcount changes and compensation numbers for the specific positions to produce a much tighter estimate that is about $40k/year less savings than the district’s number.
Lastly, SPS’s proposed Principal Office savings seem to also assume full staff removal without considering additional staff that might need to be added to the merged school. Using the same method as we did for Teaching Support, we come up with a tighter estimate for savings that produces $183k/year less savings than the district amount.
There are some smaller divergences in Grounds Maintenance costs and Utilities Cost which amount to a few thousand a year average which we leave in.
Looking at these divergences, we believe that though our estimate uses 2022–2023 data, it is based on a more granular analysis and is more accurate.
Conclusion
Summing it all up we estimate an ongoing savings of up to $630k/year to $678k/year per school assuming no major changes in enrollment get spurred by closures, and the upcoming BEX levy isn’t used to more permanently remove some of the ongoing maintenance costs in the buildings being targeted for closures. This doesn’t even reach the lower bound of the original $750k to $2.5M per building savings range that is specified in the Well-Resourced Schools FAQ.
Adding in the one-time costs of closing, the first 4 schools closed will only produce, at best, a total of $1.2M/year to $1.6M/year reduction of the district budget deficit for each of the next two years.
Furthermore, it is important to know that 72% to 78% of the savings come from staffing reductions and not building maintenance and utilities. The savings from closures are not large and they are mostly achieved by reducing the amount of staff allocated to students.
School placement should be part of long-term city planning. It has century-long impacts, similar to an urban village or transit center. Removing schools in Seattle has historically created school deserts that never get patched up and likely reinforce socioeconomic segregation as we can see the closures disparately impact vulnerable demographics. We should be discussing closures and the shift to 400+ student elementaries in the context of larger city planning, not making rapid, reactionary, and arbitrary decisions. It should include an understanding of major geographic separations like freeways, arterials, hills, and rivers, plans for creating walkable cities, and upzoning neighborhoods. Making changes in reaction to a near-term change in population is a tactical choice motivated by reasons unknown to us, not a strategic one. Especially if the savings per closure is as small as $655/year — the cost of ~4 Central Administration “Directors or Supervisors” of which we have 89 of in Central Administration with an average total compensation of $188k, or about 0.82 FTE per school and $155k/year/building of expenditure.
Where is your data from?
This analysis looks at the detailed financial data available in the OSPI F196 Child General Fund Expenditures raw data table which, since 2019, records accounting line items at a per-school level for expenditures.
Employee spending in those line items was further examined by joining it with the S275 Personnel files that report each employee’s salary, benefits, duty title, building assignments, and percentage of salary assigned to each building. A raw version of this data with obfuscated names and no demographics is available on SPS By The Numbers Bigquery as well as one joined with activity codes and building assignments. A portion of the joined obfuscated table with the District Office and the 4 targeted schools is also available in Google Sheets.
Enrollment data was taken from the OSPI Per Pupil Expenditure_AllYears dataset (PPE) which underlies the OSPI Report Card portal. The PPE data was also used to double check the numbers resulting from the F196 analysis.
Lastly, the District’s 2022–2023 Budget Book’s categorizes and summaries were used to categorize expenditures and sanity check calculations.
All data is from the 2022–2023 school year. 2024 is usually not available until December.
For transit, there is this bus funding explainer from Mary Ellen Russell.
Anyone who wants a walkthrough of the data, feel free to email sps.by.the.numbers [at] gmail.com. If you find an error, definitely reach out. We want this accurate.
Acknowledgements
This was a very difficult analysis to write up as there were so many moving parts. I want to give a special call out to Jessica Chong, and Rachel Kubiak for helping copy edit, brainstorm, and refine wording over multiple full-rewrites. Also sending thanks to Mary Ellen Russell for consulting on transit funding complexities.
Lastly, another shout out to the whole entire crew of SPS “Data Nerds” that have helped source info, clarify understandings, fact-check, math-check, sanity-check, history-check, citation-check, and in general do ridiculous amounts of work to make this as accurate as possible. All y’all rock.
And special high-five to the cadre of SPS alums who’ve reconnected in this weird fashion.
Go team!
Okay. I’m done. (though I probably forgot someone…this was a huge community effort)
