Our overall regression results, shown in Graphic 10, confirm that the city-level differences in funding, cost effectiveness and ROI persist after holding student characteristics constant.[*] On average, charter schools spend $2,611 less per student, even after controlling for student demographic differences. Charter schools are more cost effective and provide taxpayers with a higher return on investment.
On average, the charter school advantage in CAP points for each $1,000 spent does not change in any significant way when regressed against demographic factors.[†] Each dollar spent on children in Michigan charter schools provides a return on investment to taxpayers that is about $1.46 higher than the return generated by students in TPS. In other words, each dollar spent on children who persist for 13 years in public charter schools produces $1.46 more in lifetime earnings for each student than each dollar spent on children in TPS. These results are all statistically significant at the 99 percent level of confidence.
Graphic 10: Expenditures and ROI for Students Enrolled in Charter Schools for Entire Academic Career, or 13 Years
(1) | (2) | (3) | |
Per Pupil Expenditures ($) | Cost Effectiveness (CAP Points) | Return on Investment ($) | |
Charter (13 years) | -2610.755*** | 2.351*** | 1.456** |
(0.000) | (0.000) | (0.002) | |
Special Education (%) | 96.893* | -0.089** | -0.212** |
(0.011) | (0.005) | (0.007) | |
Economically Disadvantaged (%) | -15.046 | 0.004 | 0.008 |
(0.072) | (0.538) | (0.568) | |
English Language Learners (%) | 33.691* | -0.017 | 0.027 |
(0.041) | (0.140) | (0.305) | |
Minority (%) | 32.910*** | -0.024*** | -0.015 |
(0.000) | (0.000) | (0.176) | |
Female (%) | -32.903 | 0.030 | 0.040 | (0.310) | (0.476) | (0.711) |
Constant | 11776.360*** | 8.910*** | 8.734 |
(0.000) | (0.000) | (0.138) | |
Cities | 71 | 71 | 71 |
Observations | 139 | 139 | 139 |
R-Squared | 0.5407 | 0.6107 | 0.2260 |
Notes: P-values in parentheses, * p < 0.05, ** p < 0.01, *** p < 0.001
Graphic 11 below shows results for children that attend a public charter school for half of their K-12 educational experience, or for 6.5 years. On average, children experiencing 6.5 years in public charter schools have $1,470 less spent on them per year than students spending 13 years in TPS. We estimate that each dollar spent on students that persist in public charter schools for 6.5 years produces around $0.51 more in lifetime earnings than for each dollar spent on students in TPS; however, this particular finding is not statistically significant.
Graphic 11: Expenditures and ROI for Students Enrolled in Charter Schools for Half Their Academic Career, or 6.5 Years
(1) | (2) | (3) | |
Per Pupil Expenditures ($) | Cost Effectiveness (CAP Points) | Return on Investment ($) | |
Charter (Half-Time) | -1470.350*** | 1.019*** | 0.513 |
(0.000) | (0.000) | (0.201) | |
Special Education (%) | 32.018 | -0.038 | -0.105 |
(0.271) | (0.065) | (0.077) | |
Economically Disadvantaged (%) | -1.712** | 0.015* | 0.002 |
(0.008) | (0.017) | (0.887) | |
English Language Learners (%) | 31.346 | -0.015 | 0.041 |
(0.063) | (0.185) | (0.109) | |
Minority (%) | 32.565*** | -0.023*** | -0.021* |
(0.000) | (0.000) | (0.037) | |
Female (%) | -17.034 | 0.014 | 0.024 | (0.552) | (0.503) | (0.681) |
Constant | 11970.980*** | 8.633*** | 8.524* |
(0.000) | (0.000) | (0.012) | |
Cities | 71 | 71 | 71 |
Observations | 139 | 139 | 139 |
R-Squared | 0.4121 | 0.4209 | 0.1458 |
Notes: P-values in parentheses, * p < 0.05, ** p < 0.01, *** p < 0.001
The control variables behave as expected concerning statistical significance. School sectors with less advantaged student populations receive more in funding overall. School sectors with more children requiring special services have lower cost effectiveness and ROI.
The following two tables show summary statistics for the regression analysis. Graphic 12 shows descriptive statistics for the demographic variables used in the regression analysis, while Graphic 13 displays the descriptive statistics broken out for both charter schools and TPS.
Graphic 12: Descriptive Statistics of Variables Used in Regression Analysis
Variable | N | Mean | Std. Dev. | Min | Max |
ROI ($) | 142 | 8.86 | 2.87 | 2.00 | 15.89 |
Cost Effectiveness | 142 | 9.63 | 1.83 | 5.37 | 15.15 |
Expenditures ($) | 142 | 10743.14 | 2171.80 | 6194.01 | 19733.45 |
Charter | 142 | 0.5 | 0.50 | 0.00 | 1.00 |
CAP Score | 142 | 99.70 | 4.03 | 87.49 | 109.58 |
Female (%) | 139 | 49.30 | 3.17 | 38.41 | 70.09 |
ELL (%) | 139 | 6.10 | 9.10 | 0.15 | 42.00 |
SPED (%) | 139 | 12.02 | 3.83 | 1.60 | 25.74 |
Disadvantaged (%) | 139 | 53.30 | 23.48 | 7.73 | 95.10 |
Enrollment | 142 | 4639.27 | 7423.23 | 66.99 | 49392.34 |
Minority (%) | 139 | 40.50 | 30.20 | 3.83 | 99.88 |
Graphic 13: Descriptive Statistics of Variables Used in Regression Analysis by Sector
Variable | TPS | Charter |
ROI ($) | 7.89 | 9.82*** |
Cost Effectiveness | 8.44 | 10.82*** |
Expenditures ($) | 12039.72*** | 9446.56 |
CAP Score | 99.80 | 99.60 |
Female (%) | 48.31 | 50.24*** |
ELL (%) | 5.89 | 6.30 |
SPED (%) | 13.08** | 11.01 |
Disadvantaged (%) | 51.87 | 54.68 |
Enrollment | 7603.14*** | 1675.39 |
Minority (%) | 35.91 | 44.90 |
Note: * = p < 0.05, ** = p < 0.01, *** = p < 0.001.
[*] This methodology does not employ city-level fixed effects because of the small sample size. Nonetheless, all three results remain statistically significant at the p < 0.05 level when city-level fixed effects are used.
[†] Using the regression, children in charter schools gain about 2.35 more CAP points for each $1,000 spent on them than those in TPS. That compares to an unadjusted (but weighted) advantage of 2.54 CAP points for every $1,000 spent.