The story's in the data
Sometimes, you have to see the data to understand it.
These charts display the results of data analyses which form the basis of arguments in our op-ed responding to the Bridge Magazine series on school funding.
This chart shows Michigan’s average NAEP scale scores on Grade 4 reading versus the US public school average. At the same time, it shows our state rank in education spending per pupil (blue bars) and our state rank in the number of children under 18 living in poverty (tan line).
This is not a statistical analysis, but: you can see that in the late 1990s/early 2000s, Michigan was in the top ten of state per pupil spending but was also comfortably above the national average in reading test scores (though statistically tied). The state was also below the median in child poverty during this period. Starting in the mid-2000s, our spending rank started to slide, as did our test scores, falling behind the national average. Our rank in child poverty jumped up as well. The last two data points for NAEP scores are both post-COVID, where both the national average and Michigan show significant drops with Michigan falling faster. At the same time, our spending rank was barely back to the median by 2022 (last available data) and our child poverty levels were well into the top half of the country.
There is a similar pattern in Grade 4 math scores, though neither the US nor Michigan test scores dropped very significantly post-Covid. This gets much less public attention, however.
Sources: NAEP scores - NAEP Data Explorer; Michigan education spending rank - ELSI - Elementary and Secondary Information System; poverty rank - Census Bureau, Small Area Income and Poverty Estimates (SAIPE) Program
Using the same Census data Bridge and others were using, we can look at different components of Michigan spending on schools. Our state rank may put us 19th in total spending, but we’ve been lower than that on instructional spending. The biggest difference is in payments to instructional staff: thanks to MPSERS costs, we are currently 8th in employee benefits compared to other states, but we are 36th in salaries paid, even after a recent improvement. We are not putting our money in the classroom.
While US average spending on instructional salaries also dropped from 2001 to 2013, Michigan’s drop was sharper after adjusting for inflation. Michigan instructional salary spending continued to fall in real terms after 2013, until just recently, while the US average flattened out.
Sources: Annual spending - Census Bureau, Annual Survey of School Systems Finances Data (various years); state and local government price deflator - U.S. Bureau of Economic Analysis, "Table 3.9.4. Price Indexes for Government Consumption Expenditures and Gross Investment" (data originally collected from FRB St. Louis).
Fiscal 2013, which the Bridge Magazine articles used as their base year, featured one of the lowest levels of inflation-adjusted per pupil spending in the period since 2001. While Michigan was in line with the US average during this period, our state had been spending more on K-12 education in the early years of that period, and lagged behind the US average during the post-2013 recovery until FY2023.
Sources: Annual spending - Census Bureau, Annual Survey of School Systems Finances Data (various years); state and local government price deflator - U.S. Bureau of Economic Analysis, "Table 3.9.4. Price Indexes for Government Consumption Expenditures and Gross Investment" (data originally collected from FRB St. Louis).
This chart shows that the proportion of our state economy that we invest in K-12 education has fallen - in good times and bad - dramatically in the last 20 years. Even with the recent uptick, returning to our previous levels of commitment would add $6.2 billion in school aid spending.
Sources: state education revenues - Bulletin 1011 - Analysis of Michigan Public Schools Revenue and Expenditures (various years); Michigan state personal income data - U.S. Bureau of Economic Analysis, "SAINC1 State annual personal income summary: personal income, population, per capita personal income"
This chart shows both the trend in state spending on K-12 education per pupil after adjusting for inflation, and the impact of the increasing costs of paying down the MPSERS unfunded liability (UAAL). Real available funding per pupil is still down almost 11% from FY02 after adjusting for inflation. Until the big uptick in FY23, real available per-pupil funding was down by over one third (33.7%).
Sources: MI Senate Fiscal Agency, School Aid Funding History; MI Office of Retirement Systems, MPSERS Annual Comprehensive Financial Reports (various years); state and local government price deflator - U.S. Bureau of Economic Analysis, "Table 3.9.4. Price Indexes for Government Consumption Expenditures and Gross Investment".
Some perspective on what is driving higher costs of MPSERS: privatization of non-instructional services, and a shift toward charters and virtual schools, nearly none of which participate in MPSERS. The race over the last decade to save on payroll costs by contracting out non-instructional services (which do not carry the substantial cost of mandatory MPSERS contributions) has ended up undermining the system and causing costs to continue rising.
The vast majority of MPSERS unfunded accrued actuarial liability (UAAL - the estimated future cost of benefits to be paid which is not covered by estimated contributions or investment earnings in the future) comes from the system’s failure to meet (fairly optimistic) assumptions about investment income, especially in the 2010s. The system has been using more modest assumptions in recent years, but UAAL costs are still very high. This chart extends an analysis by the Citizens Research Council from 2013: Funding for Public Education: The Recent Impact of Increased MPSERS Contributions which was sparked by MIPFS testimony on the erosion of current K-12 spending from rising MPSERS UAAL costs.
Source: MI Office of Retirement Systems, MPSERS Annual Comprehensive Financial Reports (various years)
Due to changes in accounting rules for governments, pension plans like MPSERS set a schedule to pay down UAAL in a 40 year period, starting in 1998. They have kept to that schedule despite the more recent ballooning in UAAL costs. Currently, schools are paying contributions designed to bring total UAAL to zero in 2038. Costs after that reflect “normal” costs of the system.
At the same time, new hires in Michigan public schools have been put into either hybrid or full defined contribution plans, where future benefits are not guaranteed.
This is from a House Fiscal briefing from 2024. Note how the projected costs directly to districts rises until 2038, and then magically goes to near zero. It’s not clear how much policy choice is involved here. Regardless, while the fiction that the state is picking up a share of the costs is convenient, all these contributions come out of the School Aid Fund and thus reduce funding available to school operations.
Source: Michigan House Fiscal Agency, Briefing: Michigan Public School Employee's Retirement System (MPSERS) - April 2024
Most discussions of NAEP scores neglect to take into account the margins of error associated with average scores (95% confidence intervals). The NAEP program administers its tests to stratified samples of schools and districts from each state, and the margin of error increases as sample sizes get smaller. In general national data will have much narrower margins of error than state data because it is calculated based on the performance of more students. For any given sample size, however, a wider margin of error indicates that the scores in the sample varied across a wide range as well, making the simple average less useful.
In these charts, the lines for Michigan are in blue, with the margin of error for each year’s test shown as a light blue band around the line. Similarly, US national public school data is shown with a black line, with a light grey band indicating the margin of error of those average scores. Where the two margin of error bands overlap, we cannot be sure that the “true” averages for Michigan and the US are really different - if all students had taken the test under the same conditions, instead of just a sample. (The difference between the two scores is not “statistically significant.”)
For many years, especially in the 8th grade scores, Michigan was statistically tied with the US average (bands overlapping), whether that average was increasing or falling.
Note: state participation in the NAEP was not mandatory until the passage of the No Child Left Behind act in 2001, so some years are missing for Michigan and national data has wider margins of error.
